Our objective was to develop a sensitive and accurate GC-HRMS method capable of identifying and quantifying a battery contemporary insecticides in human serum. Because of recent evidence suggesting increased use and exposure to pyrethroid insecticides, the method was optimized to measure pyrethroid insecticides while including representative organophosphorus and carbamate insecticides. Because blood sampling is invasive and the amount of sample collected is typically small, it was appealing to develop not only a multi-analyte method, but one that encompassed more than one class of pesticide in order to achieve a greater exposure assessment with the limited sample volume available for analysis. However, optimizing the performance of each individual target analyte in a multi-class method is challenging when complex biological matrices are used and often the performance of some analytes must be somewhat sacrificed for the overall method performance. The diverse chemical and physical properties of the target analytes made method development and optimization challenging. Nonetheless, the method we report was the best compromise to achieve the most efficient overall extraction, cleanup and analysis of various classes of pesticides within the same 2-mL plasma sample.
Ions were selected based upon the relative abundance observed in EI spectra and the signal-to-noise ratio in the specified ion channel. For analytes that had a corresponding labeled ISTD, we also considered whether the fragment ion retained the label. Confirmation ions were selected based on relative abundance or if a naturally occurring isotope peak was present in the fragment.
Chromatographic separation was optimized to achieve separation of all analytes and to allow for analysis of individual analytes specified time segments so that optimum sensitivity could be obtained. PBUT and RM and CF and CP were the only analytes which were combined into shared time segments. For most analytes, we found no chromatographic interferences. However, small matrix peaks were present in the case of BNCB and FENV-2.
To obtain chemical behavior patterns of each analyte with a particular SPE sorbent, preliminary recovery experiments were performed by substituting the blood plasma matrix with dI-H2O. By doing so, we were able to collect information on the chemical interactions of each analyte on each SPE sorbent tested and eliminate any potential interferences or complications matrix components may present. Such information provided guidance into selecting an SPE sorbent for sample cleanup as well as the optimization of wash and elution steps for a given sorbent. Several reversed-phase sorbents (e.g. C18, C8, C2, phenyl, and CN) were evaluated for these behavioral patterns; however, we ultimately selected the Nexus cartridge for SPE because of its overall recovery efficiency and because it allowed for a greater organic wash of the sorbent which effectively minimized matrix co-extractants.
In an attempt to further minimize the presence of the blood plasma matrix in the final extract and to test the method robustness, plasma samples were prepared using the described procedure but with pH modifications of the dI-H2O and organic washes. Changes in pH might affect reversed-phase SPE retention of biological matrix components which possess various ionizable sites within their biochemical structures. Ionization of these components would thus disrupt the weak interactions between the SPE sorbent and that of the biological structure and would favor partitioning into the polar solvent and be washed away. We observed that the modifications to pH did not affect the analyte-sorbent interactions since all target analytes were non-ionizable or contained no significantly ionizable atoms at the pHs tested. No improvements or deleterious effects were observed in the analysis if at pHs of 3, 5, 7 and 9.
Chromatographic resolution of all 15 target analytes was achieved within a reasonable amount of time considering the number of compounds and the high boiling points of FENV and DM. A quantification ion and confirmation ion for each analyte and its respective ISTD was included in the measurement for increased specificity. The ions measured were selected based upon the abundance of the ion and/or the atomic composition of the fragment. In some instances, the most abundant fragment ion could not be used since the fragment ion of the ISTD was an unlabeled fragment. Use of an unlabeled fragment would result in a complete loss of specificity for that particular compound.
Most analytes were chromatographically separated and placed into individual time segments, which avoided any problems of sensitivity losses from a reduction of the accelerating voltage when the mass differences were too large. In instances when more than one analyte had to be placed in the same segment, their masses were relatively close to one another also avoiding significant reductions in the accelerating voltage. Separation allowed for increased scan times for each analyte for maximum sensitivity.
GC-injector and transfer line temperatures were set high (both at 300°C) in order to 1) vaporize the pyrethroids with high boiling points, and 2) to remove as much matrix material from the GC injector liner as possible. Still, large amounts of lipids were injected and the chromatography was negatively affected resulting in relatively large shifts in retention times over several injections. Retention shifts posed a problem for the separation of FIP and BIO making it difficult to place them into separate time segments. Separation of these two compounds was essential to maintain sensitivity for FIP. If combined into one time segment, the low masses for BIO (m/z = 123.1168 and 136.0883) and the large masses for FIP (m/z = 366.9429 and 368.9400) would result in a large reduction in the MS accelerating voltage which would ultimately result in a loss of sensitivity for FIP. The permanent presence of lipids on-column also resulted in a high background through the ion channels of BIO and PRAL which significantly increased their detection limits. Because of the varied performance of BIO and PRAL because of the lipid background, we decided to eliminate them from the quantitative method, thus their validation parameters are not reported here.
The fragmentation patterns of the pyrethroid pesticides limited the selection of ions which could be used for their analysis. The type I pyrethroids, BIO, PRAL, RM, TM, and PHEN (PM the one exception), produce extensive fragmentation of the parent molecule which typically results in a base peak at m/z = 123.1168. While the few available mass fragments proved useful as quantitation ions, the base peak ion seemed to be the only option for use as a confirmation ion. As previously mentioned, this ion channel was problematic due to the high levels of lipids present in the sample extract. Background levels were orders of magnitude higher at this mass throughout the GC run time.
The limitations of selectable ions and the presence of high levels of matrix material injected greatly affected the analysis of PHEN. shows the problems presented with PHEN. A co-eluting matrix component interferes with PHEN at its quantification ion, and appears to suppress ionization of the confirmation ion. This effect caused the confirmation ion to be unusable, and quantification of PHEN difficult. This resulted in an increase in the limit of detection for PHEN.
Mass chromatogram of phenothrin (PHEN) showing a co-eluting matrix component interfering at the quantification ion channel (A) (m/z =183.0804), and apparently suppressing ionization at the confirmation ion channel (m/z=123.1168) (B).
Many of the pyrethroid insecticides exhibited extraction recoveries that were lower than desired. This was attributed to the strong retention of these fairly non-polar compounds to the sorbent of the Nexus SPE cartridge. Recoveries were not significantly improved when using lower polarity solvents. Extraction recoveries with toluene resulted in similar recoveries as other non-polar solvents as well as increased recoveries for all other analytes. Recoveries could have also been increased with the use of another reverse-phase SPE cartridge. These low recoveries were acceptable considering the limits of detection achieved for these compounds were in the low pg/mL range. In addition, the extraction recoveries were corrected for during quantification by the use of isotopically labeled internal standards.
The LODs we report were determined statistically using the precision of repeat measurements at multiple concentrations. Therefore, the reported LODs are average estimates and do not necessarily reflect the lowest level measureable for a given analyte during a given run. In some instances, we could clearly discern a peak with a signal-to-noise ratio greater than 3 at concentrations well below the calculated LOD. However, the calculated LODs are conservative and are appropriate for average LODs over a given time period.
The RSDs and relative recoveries for each analyte were acceptable with some exceptions. The precision of the method was evaluated as within-day, between-day, and total RSDs. In general, our method presented RSD values (for all three evaluated) less than, or just above, 15%, especially for analytes possessing labeled analogues. Analytes without labeled analogues also yielded RSD values less than, or just above 15%. For analytes with higher RSDs (e.g., RM, PHEN, FENV, DM at 1000 pg/mL and CPFS, CY, c-PM, t-PM at 25 pg/mL), we found that a lack of ISTD (for FENV and RM or approaching lower concentrations near the LOD (for CPFS, CY, c-PM and t-PM), resulted in increased imprecision for those analytes. Acquisition of additional ISTDs may allow us to improve our precision for some of these analytes.
Few methods exist in the literature that have the capability of measuring the low levels of pyrethroid and OP insecticides in plasma [24
]. We previously published a method to measure a large grouping of pesticides in serum and plasma with similar LODs but only one pyrethroid insecticide was included. Methods with higher LODs allowing detection of OP pesticides following acute poisonings has also been reported [30
]. This method is highly selective but lacks sufficient LODs to be useful for general population samples. Ramesh and Ravil [31
] reported a method that measured a suite of OP pesticides in 5 mL of blood at sub pg/mL concentrations. Although their method possessed superior characteristics, our method is more selective, uses less blood, and includes OP insecticides and synergists that are also of interest.
Application of our method to archived plasma samples from a New York City cohort demonstrated that our method possessed enough sensitivity to detect cis- and trans-permethrin in a small percentage of the samples tested. Given a more acute exposure scenario, we should be able to measure low levels of these chemicals in plasma samples collected from those exposed. Notably, these samples had been stored in freezers at −80 C for up to 6 years. Because some pyrethroid insecticides have been shown to be unstable in plasma over long periods of time, we may have detected them in fewer samples because of storage biodegradation [32