Hunting the protein imposter

As melamine has been found in baby formula and soy products, fast and accurate screening of melamine and its analogues has become a necessity - Stephan Baumann tells us how

The adulteration of food with melamine – a nitrogen-rich organic chemical commonly used in plastics – has quickly become an international problem. Melamine has been detected in baby formula produced in the United States, chocolates distributed in Canada, biscuits sold in the Netherlands, condensed milk in Thailand and eggs in Hong Kong. Due to its high nitrogen content, melamine is added to foodstuffs to falsify the analysis of protein levels, making them appear higher then they actually are. Exposure to melamine-contaminated food can lead to urinary stones and renal failure1.

In response, many countries have established allowable limits for melamine. The United States Food and Drug Administration (FDA) maximum residue limit (MRL) is one part per million (ppm) for infant baby formula and 2.5ppm for other products. The FDA gas chromatography-mass spectrometer (GC-MS) screening method2 is capable of detecting melamine and its analogues (ammeline, ammelide and cyanuric acid) at 2.5ppm. However, an FDA import alert issued in February 2009 now requires that a testing method with a sensitivity of 0.25ppm for melamine and its analogues is used to assure compliance to the MRLs. Therefore, the FDA GC-MS method can no longer be used to screen for melamine and its analogues under the new regulations, and confirmation would require an additional orthogonal method. This article reports on a modification of the FDA GC-MS method that does not require a change in sample extraction and derivatisation procedures. This modification also employs a purged union GC column configuration and backflushing to provide run times under 15 minutes. With this method, melamine and its analogues can all be detected at 0.25ppm, with highly reproducible and accurate quantification and, most importantly, this approach provides screening, quantification and confirmation of melamine and its analogues all in one short run.

In this method, samples were prepared using an extraction solvent of diethylamine (DEA)/Water/Acetonitrile (10/40/50). DEA dissociates the melamine-cyanuric acid complex, reducing the risk of false negative measurements whilst improving the solubility of ammelide and ammeline, which have extremely low solubility (SP) in traditional extraction solvents. Samples were then derivatised before injection into the GC-MS system3.

To reduce chemical noise and the cycle time of the analysis, a backflushing configuration was employed to remove higher boiling substances from the column. Prior to each subsequent run, late-eluting peaks were flushed out of the inlet split flow vent instead of driving them though the entire column and into the mass spectrometer (MS). This is achieved with an Auxiliary Electronic Pneumatic Control (EPC) module or a Pneumatic Control Module (PCM) to provide a precisely controlled second source of gas to direct the column flow to the appropriate column or detector. In normal operation, the PCM pressure is at, or slightly above, the pressure of the carrier gas through the column. During backflush, the inlet pressure is dropped to 1 psi and the PCM pressure is increased, forcing the flow to reverse through the column and out the purged inlet. Such easy and rapid backflushing is enabled by the Capillary Flow Technology (CFT). CFT modules comprise a proprietary solution that allows backflushing in small dead volumes and eliminates leaks with ferrules and fittings designed for the task. Backflushing late-eluting samples out of the inlet split flow vent has the benefit of increasing sample throughput but also increases system uptime by reducing the maintenance of the columns and MS.

A unique, alternative approach to backflushing is the use of a CTM device in the middle of the analytical column4,5. Instead of using a 30m column, two 15m columns are used, and connected by an ultra-low dead volume pressure controlled tee configured with a purged ultimate union (Figure 1). The PCM adds just enough makeup gas to match that from the first column. Therefore, there is very little flow addition and subsequent decrease in sensitivity due to sub-optimal carrier gas flows into the mass spectrometer. Backflushing is accomplished, in this configuration, by reducing the flow and pressure in the first column and increasing them in the second column.

Figure 2 shows an example of backflushing with the pressure controlled tee configuration. The top chromatogram shows six standards, where the third peak is considered the last analyte of interest and the fourth peak is the first of the late-eluting interferences. The middle chromatogram shows (a) the same standard with backflushing beginning at 10.1 minutes, where flow is dropped in the first 15m column, and (b) where the flow in the second column is increased. The time between points (a) and (b) is the residence time of the last analyte compound in the second column. With this method, you retain the last analyte, but the late eluters never enter the MS. The bottom chromatogram demonstrates the lack of carryover in a subsequent blank run. Alternatively, backflushing can begin after the last peak of interest has eluted (point b). This eliminates the need to determine experimentally the residence time of the last target compound in the second column, while slightly increasing the cycle time.

This method, developed on a GC/MS/MS system, provides excellent separation and analysis of melamine, ammelide, ammeline and cyanuric acid in a single run, and in less than 15 minutes (Figure 3). There is significant improvement in the sensitivity and selectivity of this new method versus the GC/MS SIM (selective ion monitoring) method. The GC/MS SIM method is less effective at reducing chemical noise at 2.5ppm, using any of the SIM ions, whilst the new method provides a very clean analysis of the quantifying transition of melamine at 0.25ppm.

Each of the standards for melamine and its three analogues was added to a matrix of either baby formula or soy meal at concentrations of 0.78, 1.25, 3.9 and 12.5ng/ml, corresponding to detection levels of 0.16 to 2.5 ppm. Calibration curves constructed for each of the four compounds in each matrix3 (data not shown) demonstrated excellent linearity obtained for melamine and its three analogues, which is demonstrated by having R2 values very close to 1.00. The accuracy of quantification was also very good for all four compounds in both matrices.

The identification point system, developed by European Union scientists, defines an acceptable procedure for scientifically confirming the presence of regulated substances6. The more identification points provided by the analytical method, the more certain is the confirmation of the compound. Three points are required for compounds with a minimal reporting level (MRL). In cases where the toxicity of the compound means you cannot define the MRL, the compound is banned at all levels. Four identification points are required to identify an analyte7. With GC/MS, four ions are monitored to provide four identification points but when using the GC/MS/MS only two selected reaction monitoring (SRM) transitions need to be monitored. Analysis of melamine and its analogues on this GC/MS system used at least two SRM transitions for each compound to provide screening and positive confirmation in the same run.

Figure 4 illustrates the quantifying and qualifying transition profiles for the GC separation of each of the four compounds in both baby formula and soy meal. In each case, the qualifying transitions have been normalised to the quantifying transition in order to better illustrate the identical peak shape obtained from both. These transitions therefore provide a positive confirmation of each of the four compounds in each of the sample matrices.

In conclusion, this modification of the FDA GC/MS method for screening for melamine, ammelide, ammeline and cyanuric acid provides screening, quantification and confirmation in one short run. This method does not require any changes in extraction or derivatisation procedures, and the cycle time is about 15 minutes. In addition, this method meets the new FDA requirement for sensitivity of 0.25 ppm, and it demonstrates excellent linearity of quantification up to 2.5 ppm. Using this approach two SRM transitions for each of the four compounds provides sufficient identification points for a positive confirmation with very good accuracy of quantification.

Acknowledgement
The author gratefully acknowledges the guidance and material assistance of Greg Mercer, Pacific Regional Laboratory Northwest, Food and Drug Administration.