Issue Date: November 30, 2009
Testing Life's Staples
“Laboratory testing is an essential component of a science-based food safety system,” said Food & Drug Administration Commissioner Margaret A. Hamburg during congressional testimony in June. Food safety bills now wending their way through Congress would, in part, increase laboratory testing as a way to prevent food-borne illnesses.
Laboratory testing for food safety, however, is complicated. Analysts must watch for contaminants as varied as bacteria and pesticides in matrices from burritos to peaches. To help manage it all, researchers and instrument companies are working on ways to streamline, automate, and accelerate sample preparation and analysis, as well as increase sensitivity to cope with new regulatory limits.
To start, there is the vast number of things to look for—chemicals such as pesticides, veterinary drugs, mycotoxins, and harvesting or processing contaminants, as well as biological hazards such as bacteria, molds, viruses, parasites, and allergens. And then, as incidents of drug and food adulteration have illustrated in the past few years, there are the things that food analysts don’t know to look for.
There is also a wide variety of matrices at play, from raw fruits and vegetables to processed foods such as cereal or canned stew. Challenges exist in sampling, sample preparation, and actual analysis—especially if food producers and manufacturers also have to watch for unanticipated adulterants in addition to known contaminants. Last but not least, as analysts cope with all of the above, they’re under intense pressure to produce results quickly enough so that the food in question doesn’t go bad while they are running the tests.
Sampling is where food analysis begins. Sampling a crop for pesticide residues is fairly straightforward because agricultural chemicals are typically applied uniformly to a crop, says Trevor Suslow, a researcher in the Postharvest Technology Research & Information Center at the University of California, Davis.
Biological contaminants, on the other hand, are generally not homogeneous. Escherichia coli bacteria from wildlife feces, for example, are apt to take hold in a localized hot spot. “You’re probably more likely to miss it than to find it,” Suslow says, noting that the patchy nature of pathogen colonies is currently one of the biggest obstacles to ensuring the safety of fruits and vegetables.
One solution would be a product wash step, from which liquid could be collected, concentrated, and analyzed. Another would be in-line spectral analysis that could check each unit of food as it passes by on a conveyor belt. “But these are not trivial solutions,” Suslow says, pointing out the high cost of purchasing, installing, and maintaining equipment.
Once a product has been sampled, it might go through some sort of in-field rapid screening. U.S. Department of Agriculture meat inspectors, for example, use a field test for bacterial growth to check for broad-spectrum antibiotic drugs in cattle kidneys, says Emilio Esteban, USDA science adviser for laboratory services and research. An inspector cuts a notch into an animal kidney, swabs the organ, and then inserts the swab into a tube inoculated with bacteria. If no antibiotics are present, a six-hour incubation period allows those bacteria to grow and produce acid; the drop in pH causes a colorimetric indicator to change from purple to yellow. If antibiotics are present and kill the bacteria, however, the tube stays purple.
Field tests, though, tend to be nonspecific. Generally, all that current technology can do is raise a red flag that further analysis is needed. In the case of a positive kidney test, USDA then must send kidney samples back to an agency lab for further testing to determine the identity and amount of the drug or drugs that are present.
And sending a field sample to a lab for analysis takes a significant amount of time. “People are dreaming” about handheld instruments that could do everything in one spot, whether in an agricultural field, in a meat-packing plant, or at a port, says Stuart Cram, vice president for strategic marketing of scientific instruments at Thermo Fisher Scientific. The goal would be something that is low cost but that has the resolution of a mass spectrometer. The Department of Homeland Security is working on ways to make their chemical labs more mobile, Cram says, and that technology could eventually be brought to the food industry.
For now, however, samples in need of detailed analysis must go to a lab, where the next step is to prepare the sample for testing. Analytical instruments that could handle a simple “dilute and shoot” approach would be ideal, “but that requires greater sensitivity on the part of the instrument,” says Joe Anacleto, vice president of applied markets and clinical research for Life Technologies’ mass spectrometry division. “A lot of sample prep involves doing some concentration of your sample so that you can detect lower levels.”
The standard sample prep technique to extract and concentrate pesticides, antimicrobials, and other chemical contaminants is based on a method known as QuEChERS, which stands for quick, easy, cheap, effective, rugged, and safe. The technique was developed earlier this decade by USDA scientist Steven J. Lehotay and colleagues (J. AOAC Int. 2003, 86, 412).
The general approach involves extracting of a frozen 10-g sample in a solvent such as acetonitrile, followed by liquid-liquid partitioning. A dispersive solid-phase extraction using a polymer sorbent then removes residual water and cleans the sample by removing polar matrix components, such as organic acids and sugars. The resulting sample can be analyzed by either gas or liquid chromatography.
Lehotay and colleagues have continued to work on QuEChERS-based sample preparation methods that enable broad multiresidue screens, USDA’s Esteban says. The USDA Food Safety & Inspection Service alone receives and tests roughly 150,000 samples per year for chemical and microbial contaminants. “We can’t just take one sample at a time and analyze for one thing, we need to analyze for a lot of things,” Esteban says. To accomplish that, USDA tries to limit sample preparation to four or five methods that can be used to target compound classes, such as herbicides and pesticides or antimicrobial agents.
But those methods could still have to be modified depending on the sample matrix. By one estimate there are more than 50,000 different matrices of foods, beverages, and agricultural products, Thermo Fisher’s Cram says. Coming up with a sample prep method for each of those requires “working backward from the regulatory detection level, data-quality requirements, and chemical background that you’re likely to encounter,” he says. “Then you have to focus on the nature of the sample, whether it’s ice cream, fruits and vegetables, or calf liver.”
The time and labor required to isolate contaminants from the sample matrix is the rate-limiting step for food safety analysis. One newer approach to automating and accelerating sample prep is turbulent flow chromatography. This method involves using a column packed with large particles—40–50 μm in diameter versus the 2–5-μm-diameter particles in a typical high-performance liquid chromatography column—and forcing the mobile phase through the column so quickly that eddy currents develop. The combination of large particles and eddy currents means that larger molecules flow through the column faster than smaller molecules. A sample need only be extracted and centrifuged (if it is a liquid, only centrifuged) before being injected directly onto the column for cleanup, after which it can be directly routed to an LC or LC/MS system.
Another new technique for sample prep is solid-phase microextraction, in which a sorbent-coated fiber absorbs either an analyte from a liquid or, in the case of volatile compounds, the vapor phase from a vial’s headspace. There is still an equilibration time, notes Yolanda Fintschenko, manager of food safety technologies at Thermo Fisher, but the process can be automated. Overall, solid-phase microextraction uses little solvent, making it one of the more environmentally friendly approaches to sample preparation.
Once a sample is prepared, the next hurdle is the actual analytical testing. Testing all foods for all possible contaminants, whether chemical or biological, is cost- and time-prohibitive.
“We design the quality control we should do on different products depending on the risk of finding particular contaminants in specific matrices,” says Fabien Robert, who heads the Compound Identification Group in the Department of Quality & Safety at the Nestlé Research Center. The company will typically check cereal grains for mycotoxins, for example, and fruits and vegetables for pesticides.
To check for chemical contaminants or for quality control in the field, Nestlé primarily uses methods such as immunoaffinity assays. In factories, the company uses techniques such as near-infrared spectroscopy to screen raw materials or semifinished and finished products, Robert says. Near-IR spectroscopy can detect contaminants down to about parts-per-million levels; if the company is concerned about lower concentrations, it turns to more sensitive and specific techniques, such as HPLC with ultraviolet detection. If one of the rapid testing methods sends up a red flag, the company then sends samples to one of its specialized labs, where analysts typically turn to MS methods to identify the specific compound involved.
As regulations tighten and contaminants such as residual pesticides and antibiotics have to be held at lower levels, MS is increasingly moving to the front line of food analysis, says David J. Elliott, president of contract lab Environmental Micro Analysis.
Environmental Micro Analysis’ choice of technique depends primarily on two things. First, the analysts look at the chemical properties of the contaminant. Elliott notes, for example, that many newer pesticides, such as Bayer CropScience’s spirotetramat for aphids, cicadas, and other sucking insects, are not vapor-phase stable and cannot be analyzed by GC. Second, the analysts consider the regulatory requirements. If the contaminants are limited to the parts-per-billion range, MS—and, in particular, a tandem MS setup—often becomes necessary.
More generally, labs of all kinds are working on multiresidue analysis methods that would allow a single sample to be run through several MS detectors, enabling analysts to screen for 200 or more compounds in one shot.
On the biological contaminants side, the analytical workhorses are immunoaffinity assays and polymerase chain reaction (PCR)-based tests. The challenge in this area is that FDA guidelines say that tests need to detect as few as one or two bacteria in 25 g of meat. Doing that directly “is out of reach of any of the current capabilities in any kind of system today,” says Manohar Furtado, a scientific fellow at Life Technologies. What analysts do now to achieve FDA’s detection limit, as well as to determine whether an organism is actually alive, is to grow some sort of bacterial culture.
Culture growth is far and away the biggest time sink in bacterial identification. With the usable life of perishable foods hanging in the balance, food companies are anxious to find ways to reduce the time it takes to achieve results, Furtado says. Using larger amounts of culture broth, richer media, and magnetic, antibody-laced beads can lead to E. coli identification in less than eight hours, compared with as long as 27 hours.
Other approaches are real-time PCR-based tests that amplify genetic material and detect sequences of interest within a single tube so that DNA doesn’t have to be removed and run on a gel. Life Technologies is also looking at using RNA instead of DNA targets, Furtado says. Although an organism might have only one or two copies of a particular gene target, it could have 50–60 copies of the analogous RNA, making it more likely that the nucleic acid sequence in question will be detected.
USDA uses PCR-based tests to look for bacteria, in particular E. coli, Salmonella, and Listeria, and plans to add Campylobacter to its regular screens in 2010, Esteban says. Once a sample arrives in the lab, agency scientists can have the result in 24 hours. But it’s not enough to determine whether bacteria are present, he adds. In the case of bacterial pathogens, the genus, species, and particular strain must be identified. Most E. coli, for example, are harmless, but the O157:H7 strain can cause bloody diarrhea and kidney failure and has been implicated in several U.S. food recalls.
One emerging biological contaminant causing food safety concerns is the allergen. FDA analyzes foods to look for possible allergens, such as particular proteins in peanuts, as part of surveillance to ensure labeling accuracy. In this case, analysts are looking for a very small amount of an allergenic protein in a matrix that, of course, is full of many other proteins, says Steven Musser, who directs the Office of Regulatory Science at FDA’s Center for Food Safety & Applied Nutrition.
Allergenic proteins can be present in the low parts-per-million to high parts-per-billion range, “so getting them out and proving that they’re there is an emerging area,” Musser says. “We don’t yet have a QuEChERS equivalent for proteins.” A starting approach for something like peanut allergens in chocolate would likely involve defatting the sample, purifying the protein, enzymatically digesting the protein, and then using LC/MS techniques to identify the resulting peptides. That sort of analytical approach “is very common in the clinical proteomics field,” Musser says, but on the sample prep side, foods are much more challenging than the typical clinical samples of urine or blood.
Still, all these challenges pale in comparison with the biggest threat to modern-day food production: the unexpected. And that could be a terrorist adding cyanide or ricin to the food supply or an unethical supplier substituting a cheaper—and possibly toxic—ingredient.
“The biggest thing that is scaring the heck out of people is looking for things they don’t expect, like melamine,” says Paul Zavitsanos, Agilent’s worldwide food industry manager. “Now people are asking, ‘How do we not leave these things to chance? What kind of analysis workflow and systems can we use to catch these things at the first pass?’ ”
Starting approaches include using any analytical technique, from high-resolution MS or nuclear magnetic resonance spectroscopy to near-IR spectroscopy, to acquire as much data as possible. Then analysts can apply software and chemometrics to mine the data for things that don’t fit a typical fingerprint. But “you have to be sure that you have well defined the space of your model,” Nestlé’s Robert says. A model that is too narrow will yield many false positives. But if it isn’t narrow enough, it risks not picking up an adulterant or contaminant. “The challenge is to find the right balance,” Robert says, noting that Nestlé is also trying to anticipate why and which food ingredients would or could be adulterated and to develop specific methods to address those concerns.
The challenges facing analytical chemists guarding our food are relentless. Even as their methods become faster and more sensitive and new ones are developed, expanding global sourcing of food, identification of new threats to safety, and implementation of new regulations will continue to put pressure on them. Their efforts help ensure that we can sit down to a safe dinner each night.
- Chemical & Engineering News
- ISSN 0009-2347
- Copyright © American Chemical Society