The way forward for FAMEs analysis?

When it comes to automating sample preparation for FAME analysis - can robots really replace the proven expertise of a technician? Ray Perkins tells us why he thinks they can…

The composition of triglycerides is of great importance to scientists working in fields such as, medical clinical and nutritional research, crop science, food science and industrial chemistry and the determination of the fatty acid composition of triglycerides is a common requirement. It is usual for lipid sample to be prepared for analysis, by forming fatty acid methyl esters from the triglycerides. A boron triflouride / methanol mixture is often used as a derivatisation reagent, with analysis if the resulting mixture by gas chromatography. This sample preparation process is typically carried out by hand, however this article shows that it feasible to implement automated analogues of these manual procedures and a comparison is made between automated and manual variation of one such procedure. The study shows that automation of the sample preparation is viable and gives results that are at least as good as the manual version of the method.

When an analytical laboratory is faced with a large increase in the number of samples to be run; the economics of the analysis changes significantly. From a cost accounting perspective, the costs associated with any analysis can be seen as composed of two elements: variable costs (those costs that rise in proportion to the increase in samples analysed) and fixed costs (those costs that remain the same as the number of samples analysed increases). If the aim is to minimise the total cost of the activity, when sample numbers are low, it is relatively important to minimise fixed costs, since these costs will tend to represent the major component of the total cost. However, as sample numbers rise, the proportion of the total cost represented by the fixed costs will fall sharply, the reverse argument will become increasingly true and so it will become relatively more important to minimise variable costs (even at the expense of increasing fixed costs). Labour costs associated with sample preparation are usually a large part of the variable cost component of any analysis. As a consequence, when faced with an increasing sample workload, automation of the process will be a sensible  option when the savings in the labour component, out-weigh the increase in fixed cost due to the capital spend required.

Once automation becomes economically favourable, increased automation offers other significant advantages such as opportunities to improve the precision of the analysis by reducing the effect of human variability, reduce exposure of laboratory staff to hazardous materials and to minimise the use of environmentally undesirable substances.

Triglycerides are the main components of fats and oils from animals and plants. The commercial and dietary significance of these substances makes the analysis of these compounds a common one.  Triglycerides are made by the combination of three fatty acid molecules with a single glycerol molecule, by the formation of ester linkages between the three OH groups of the glycerol molecule and individual fatty acids. The most common methods used in their analysis involve breaking the ester linkages, forming the methyl esters of the fatty acids and analysing the mixture of fatty acid methyl esters (FAMEs) to determine the fatty acid composition of the fat or oil.

Although the fundamental approach to this chemistry is the same for laboratories performing lipid analysis in widely different contexts, the details of the methods vary widely from one laboratory to another. In many cases, the analysis has been performed for many years and the reasons behind the choice of the parameters used and the extent to which they have been optimised are no longer known. Temperatures, incubation times and reagent and sample volumes show large variations from one laboratory to another. The decision to automate the derivatisation provides an excellent opportunity to explore the details of individual procedures thoroughly and to determine and document an optimum parameter set that is suitable for automation.

Beyond the derivatisation itself, it may well prove feasible to incorporate into the method other manual steps in the analysis, such as the addition of internal standards, evaporation and sedimentation steps.

The general approach to automation detailed here has been show to work successfully for procedures used for the determination of the fatty acid composition of industrial oils, blood serum and blood-spot samples. The example that follows is one example of the successful automation of a routine FAMEs analysis, that serves to illustrate the feasibility and practical benefits of automated FAMEs derivatisation.

This example describes work done to automate the preparation of fatty acid methyl esters from lipid samples and deuterated fatty acid surrogates, prior to analysis by gas chromatography and includes a comparison of results obtained automatically with results obtained using the existing manual procedure. The lipids samples in this example were derived from polymer specimens via accelerated solvent extraction (ASE), followed by evaporation of the extract to dryness. Experimental The use of boron triflouride and methanol for the preparation of fatty acid methyl esters (FAMEs) from lipids is a commonly used and well documented procedure1. FAMEs are non-polar and more volatile than their corresponding fatty acids; therefore they are much more amenable to analysis by gas chromatography than free fatty acids. In this study, to maintain comparability, an automated version of the procedure was devised to mimic the manual version as closely as possible.

Samples were prepared on a “just-in-time” basis to ensure that all samples would be treated in the same fashion. The lipid samples were presented in the form of a dried extracts in a 10ml vials. A GERSTEL MPS PrepStation, set to accommodate 10ml sample vials, was configured with 10µl and 1ml syringes, dual heated agitators, and a four place solvent delivery station. Two of the solvent delivery stations were filled, one with HPLC grade water, and one with acetone (used as a co-solvent for rinsing of the syringes). Two separate 100ml vials were used to contain the hexane/internal standard mixture (bromotetradecane in hexane) and the derivatisation reagent (14% BF3 in methanol) Programming and control was via GERSTEL Maestro software. The processed samples were analysed using an Agilent 6890 GC with 5973 Mass Selective Detector and 7673A auto sampler. The configuration of instrumentation arrived at, allowed for the entire sample preparation, injection and analysis to run as a seamless process or for the sample preparation and analysis to run independently, whichever mode of operation best suited the workload of the laboratory.

Table 1: Areas for Methyl Esters from Manually Derivatised Extracts 2

For the purposes of this comparative exercise, the process was halted after step 6 and a portion of the organic layer removed with a Pasteur pipette and sealed in a 2ml auto sampler vial, for subsequent analysis with a parallel set of samples that had been prepared manually by a skilled and experienced technician. Since the time taken to process a single sample was significantly greater than the cycle time of the gas chromatograph, the fact that the robot processed the samples were in a serial fashion meant that in order to gain maximum productivity from the GC-MS, it was important to ensure that the preparation of two samples was in progress at any one time. To this end, the PrepStation was used in “prep ahead” mode, which triggered the Maestro software to interleave the processing of each pair or samples to ensure that samples were processed at a rate that matched the rate that the GC could run samples. This approach is made possible in this example, by the fact that the process involves two lengthy mixing stages during which time the robot is free to begin processing the next sample in the sequence.

Table 1 contains data from the manually processed control samples and Table 2 contains data from the extracts produced by the PrepStation. Table 3 contains summarised results from Tables 1 and 2. The variability in the data for the target endogenous methyl esters include a contribution from the accelerated solvent extractions (ASE) which was performed individually on each sample, prior to submission for derivatisation. Known concentrations of the deuterated acids were spiked into the extracts post ASE and prior to drying ready for derivatisation. The derivatised extracts prepared by each technique were alternated within the GC-MS sequence to compensate for potential drift in the performance of the GC-MS throughout the sequence of samples.

Peak Areas for Methyl Esters from Automatically Derivatised Extracts2

These results prove that it is possible to use a sample preparation robot to successfully mimic a common manual process used for the preparation of fatty acid methyl esters. The results obtained using automated sample preparation compare well with results obtained when a skilled and experienced technician prepared identical samples by hand. As expected, the automatically derivatised samples demonstrate better precision than the manually prepared samples and, in this case, recoveries were also found to be slightly better. Using the robot, it is easy to arrange the sample processing sequence such that several samples can be worked on at any one time while maintaining a regime where each sample is processed in an identical fashion and is also completed just as the GC comes ready to run each sample, something that even the most experienced technician would find difficult to achieve. Automation was shown to be a practical alternative to the manual preparation of FAMEs that offered a significant cost saving by freeing valuable staff from a time consuming routine task to spend time on more cerebral activities.