Methods to determine the extent of lipid oxidation in foods - Oxidation in Foods and Beverages and Antioxidant Applications - page 189

siloxane), is immersed in a liquid or gas sample causing partitioning of com-
pounds between the sample and the fiber surface. The adsorbed compounds are
then thermally desorbed in the GC injection port to enter the column for
separation (Yang and Peppard, 1994). SPME has the same sensitivity as
dynamic headspace but is less expensive and have better precision and accuracy
(Marsili, 1999).
8.5 Sensory analysis and correlation between sensory and
instrumental analyses
While there are a number of options to study oxidation in pure lipids and lipid
extracts, evaluation of the degree of alterations in complex foods at low
oxidation level is still a tricky subject. No analytical method has, to date, proved
reliable in describing low degrees of oxidation in complex foods, typical
examples being off-flavor in milk/milk powders or warmed off-flavor in meats.
Sensory analysis of volatile compounds is the method of choice in most of these
and other complex cases but it suffers from difficulties and high costs of
performance. Sensory assessment of lipid oxidation is generally performed by
trained panels, who assess flavors such as grassy, painty, rancid, buttery, fishy,
beany, etc., commonly on a 10-cm flavor score 10-cm line scale ranging from
`not detectable' to `intense'.
Comparisons between sensory and instrumental analyses have been routinely
performed in studies, and good agreements were generally found between the
sensory analysis and the levels of secondary oxidation products in oxidized lipid
mixtures. Good correlations were, for example, obtained between sensory
descriptors, TBARS, and volatile compounds (Kulshrestha and Rhee, 1996;
Salih et al., 1987; Shahidi and Pegg, 1994; Tikk et al., 2008). Sensory
descriptors of rancid odor (typically oxidized, painty, and grassy) correlate with
volatile oxidation products such as hexanal, 2-hexanol, heptanol, 2-octenal, and
2,4-decadienal and TBARS correlate better with sensory scores than PV
(Maisuthisakul et al., 2007). Correlations between grassy and painty scores and
volatile oxidation products in whole milk powder (Lloyd et al., 2009) are given
as examples:
Grassy score 0.92 0.00058*hexanal 0.024*2-heptanone
0.073*1-octen-3-ol ÿ 0.083*nonanal
(R 2 0.38, p < 0.0001), and
Painty score 0.47 0.00073*hexanal 0.018*Octanal
0.07*2-methylbutanal ÿ 0.086*3-methylbutanal ÿ
0.022*2-octen-2-one (R 2 0.61, p < 0.0001)
Sometimes, the correlations are not linear. For example, an exponential
relationship was found between sensory score and log TBARS in cooked pork
(Nissen et al., 2004).