OR WAIT null SECS
Jim Garvey of the Pesticide Control Laboratory in Ireland discusses the development of pesticide analysis methods with high-resolution accurate mass MS.
The analysis of pesticides and other contaminants in food presents ongoing challenges, particularly for laboratories conducting routine testing. Updated regulatory requirements require testing for expanded lists of pesticides and at ever lower detection levels. Multiresidue methods assist such laboratories in handling demands for high throughput, but detecting ever-lower maximum residue limits can be difficult with such methods. One approach to address these demands is to develop methods that use high-resolution accurate mass (HRAM) mass spectrometry (MS). Jim Garvey, the Head of Food Chemistry at the Department of Agriculture, Food, and the Marine, in Celbridge, Ireland, recently spoke to us about how his group develops methods, including those using HRAM-MS, to meet ever-changing demands in pesticide analysis.
What are the biggest concerns you and your organization face in your work regulating the use of pesticides in Ireland?
The Pesticide Control Laboratory is the national monitoring laboratory for Ireland, and one of our primary functions is the implementation of European Union (EU) law within Ireland. This means that each year we face an increasing list of pesticides and metabolites to be included in our scope. This in turn means that we need to have a strong method development function. Added to this, the “race to the bottom” is never far away: Legislators look at the increasing sensitivity of instrumentation, and ask if this means EU maximum residue levels (MRLs) should be lowered. The current default value is 10 parts per billion (ppb), which is achievable for most pesticides, at the moment.
Monitoring laboratories have a different approach to this increase in instrument sensitivity: We use it to dilute samples to minimize matrix effects, particularly in liquid chromatography–mass spectrometry (LC–MS) methods. Reducing the MRLs would mean that this option would no longer be available to us, and matrix effects would again become a major problem.
Probably the major concern at the moment is the new legislation covering monitoring laboratories, EU 625/2017 (1). In the past, it was enough for laboratories to be accredited, but now this has been expanded to cover all methods and their scopes. In the pesticides area, we use multiresidue methods, which can cover scopes of up to 800 analytes. There is an acceptance that not all of these analytes will behave perfectly. So, we are faced with the prospect that we may have a lot of data that we cannot report. Potentially this means we will have to develop a lot more single-residue methods, which will eat into the efficiencies we have developed in the past decade.
What are the biggest challenges you face overall as you work to address those concerns and to achieve your mission?
A strong method development function is critical. Our aim would be to cover more than 90% of the pesticides and metabolites using multiresidue methods. This means that our current methods are constantly expanding; for example, we have a plan to add another 45 pesticides to our existing methods by the end of 2019. To cope with this, we have had to look at alternatives to our existing triple-quadrupole MS systems and different workflows such as the introduction of screening methods. For this reason, we have started to develop and validate methods using high-resolution accurate mass (HRAM)-MS systems.
What are the biggest challenges you face in terms of method development for pesticide residue methods?
Probably the biggest challenge we face at the moment is the anionic pesticides, such as glyphosate, glufosinate, ethephon, fosetyl aluminium, and so on. These are covered by the quick method for polar pesticides (QuPPE) within the EU. In the case of glyphosate, derivatization was originally required, and then many column types were tried including hydrophilic interaction chromatography (HILIC), multimode, porous graphitic carbon, and ion exchange, but we are still struggling to find a robust and reliable method. A lot of people are now looking at ion-exchange chromatography coupled to MS as the solution to this problem, but these methods are still under development.
There is also a range of compounds that have been around a long time that are still causing us problems, such as captan, captafol, dicofol, Folpet, and binapacryl. These are problematic for different reasons; the hard ionization of electron impact (EI), degradation, liner effects, and so on. We are looking at atmospheric pressure gas chromatography (APGC) as a possible method for these compounds, and our initial results have been quite encouraging.
Then there are the difficult matrices such as spices, tea, and coffee, which provide us with serious sample preparation problems.
You have developed a multiresidue method for the analysis of pesticides and polychlorinated biphenyls (PCBs) in fruit and vegetables using GC–HRAM-MS. First, why do you feel it is necessary or worthwhile to move to HRAM-MS for this type of analysis, rather than continue with triple-quadrupole MS?
We have used triple-quadrupole instruments now for almost 15 years, and, in fairness, they have served us well. The demand for an increase in scope means that we will see methods with scopes of 1000 pesticides and metabolites or more in the near future. We may also be asked to combine methods for pesticides and other residues. This has already happened for mycotoxins and is starting for veterinary drugs. I think methods using triple-quadrupole instruments will struggle to cope with this demand for a number of reasons. We need to find transitions for every new molecule added to a method, and these transitions need to be optimized for the instrument being used, which is a considerable amount of work. Once we have done this, the scanning speed of the instruments then limits the number of transitions that we can fit into the method, and this in turn puts a limit on how far we can develop triple-quadrupole methods. These limitations don’t exist with high-resolution MS systems, and these systems also give the advantage of high resolution and mass accuracy, which gives us greater confidence in our results.
What is the sample preparation for this method? Did you need to change that from what you were doing previously?
No. The sample preparation method is the Dutch miniLuke method, so this hasn’t changed, at least not for fruit and vegetables. Although the QuEChERS (quick, easy, cheap, effective, rugged, and safe) method is more popular with pesticide laboratories generally, we find that the Dutch miniLuke method is a robust and reliable method, and is applicable without modification to a wide range of matrices.
Can you briefly describe the GC–HRAM-MS method?
The method uses a TG-SilMS, 0.25 mm x 0.25 μm column (ThermoFisher Scientific) with a 5-m safeguard. We use a programmed temperature vaporizing (PTV) injector with a temperature ramp and a 1-µL injection volume. A pretty straightforward temperature program is used. The MS instrument is a QExactive Orbitrap (ThermoFisher Scientific) with EI ionization, and we use it in full scan mode with a resolution of 60,000. From data collected, we use one ion for quantitation and for most pesticides three confirmatory ions (Figure 1).
What were the biggest challenges in developing this method, as you move from using a triple-quadrupole system to using HRAM MS?
It’s a different way of thinking, and the data processing is quite different. You become a little dependent on software, and, therefore, the reliability of the software is critical. We need to make sure that the criteria for peak finding are robust. We need to have the confidence that the software will not fail to integrate a peak for any reason, such as interference or peak movement. To build this confidence, we ran the two systems in parallel for a period of time and compared the results. The agreement was excellent. One of the main challenges is the vast amount of data collected, which then needs to be evaluated and requires a significant increase in storage space.
How did you validate the method?
The validation protocol we used is a combination of the criteria in the SANTE/2017/11813 guidance document (1,2) used in the EU with additional criteria required by our accreditation body. The method was for fruit and vegetables, so we used a high-water content matrix (cucumber), a high acid content matrix (lemon), and a high chlorophyll content matrix (broccoli). We carried out recovery experiments across the linear range of the method (5–250 ppb) using at least two different analysts, and we calculated the repeatability and the within-laboratory reproducibility. Mass accuracy and the confirmatory ion ratios were also evaluated, and lastly we looked at matrix effects.
What results have you been able to achieve with this method, in terms of recovery levels, precision, mass accuracy, reproducibility, and repeatability? How do those results compare to what you could achieve with the previous method?
The method contains 167 pesticides, PCBs, and metabolites, and was successfully validated for 94% of these. Linearity is very good for the vast majority of analytes. This means that our recoveries are within 60–140% with a repeatability and within-laboratory reproducibility of <20% (Figure 2). The sensitivity in full scan mode easily meets the default MRLs of 10 ppb for 90% of the analytes. The mass accuracy for the target ion and confirmatory ions is less than 2 parts per million (ppm) for most of the analytes. Although it’s only an indicative criterion, we evaluated the ratios of confirmatory ion to target ion, and again the majority of these meet the 30% level set in the SANTE document. Matrix effects were evaluated, but we tend to use matrix-matched standards to minimize these effects. Finally, the method was tested in proficiency tests and the results were excellent. Compared to our existing triple-quadrupole methods, the selectivity is much better, the sensitivity is at least comparable and in a lot of cases better, and the repeatability and reproducibility are better.
Did some of the compounds you analyzed present more challenges than others?
Yes, with GC the usual suspects caused problems: captan, Folpet, captafol azinphos methyl, dimethoate, omethoate, phorate, and dichlorodiphenyltrichloroethane (DDT).
How did you deal with matrix effects?
We use matrix-matched standards. For validation, we use an exact matrix match but when we run samples, we don’t have this luxury, as a normal batch can contain a variety of matrices, so generally the match is not exact.
Can this method be used for quantitation?
Absolutely. The method is validated for quantitation across the linear range of 5–250 ppb. You can see an example of the analysis of azoxystrobin in cucumber in Figure 3. Results outside this range that require dilution need to be accompanied by additional recovery work.
For laboratories working in routine, high-throughput analysis of food samples, is HRAM a good choice? Would you anticipate challenges for such laboratories, such as lack of access to this instrumentation or lack of experience using it?
The Pesticide Control Laboratory is such a laboratory, and for us it is a good choice. The challenges are lack of experience and evaluation of the huge amount of data collected compared to methods using triple-quadrupole instruments. Another important consideration is price, because HRAM systems are currently significantly more expensive than the triple-quadrupole systems.
There is increasing use of data-independent acquisition (DIA) in environmental analysis. Are you seeing this trend in food analysis as well? Are you pursuing the use of that approach?
For GC, this approach is unnecessary, as we get sufficient fragmentation in electron ionization (EI) mode and full scan experiments work very well. We use this approach for LC, where electrospray ionization (ESI) is a much softer fragmentation process, and secondary ionization is necessary in most cases. The main advantage of the DIA technique is that it minimizes matrix interference. There are usually two areas of criticism of this technique: that there is no physical connection between the initial full-scan experiment and the DIA experiment and this means the judgment of the analyst is a critical factor, and that the technique is heavy on cycle time. These criticisms are true, but as long as we keep the cycle time below 1 s, then we can collect sufficient points across the peak to give accurate quantitation.
You have also been using high-performance ion chromatography in your work. What is driving that move?
The anionic pesticides are one of the biggest challenges we have at the moment, especially with the recent controversy surrounding glyphosate. The pesticides have proved difficult to analyze, and the methods used all have issues. I feel that ion chromatography has the potential to give us the most reliable methodology for the analysis of these compounds.
What stage are you at in that work?
We are in the process of validating this method for cereals. Once that is finished, then we will move on to fruit and vegetables.
What are your next steps in your work in pesticide analysis overall?
We have a long list of projects at the Pesticide Control Laboratory. The main priorities at the moment are to validate the LC–HRAM method for fruit and vegetables and the IC–HRAM method for anionic pesticides. After that we move on to other matrices such as cereals, milk, eggs, and infant formula. In parallel with this, we would like to test and implement targeted screening methods for these matrices.
Jim Garvey is the Laboratory Manager at The Pesticide Control Laboratory at the Department of Agriculture, Food, and the Marine, in Celbridge, Ireland.