Sensory characteristics, such as appearance, taste, and aroma, are the basic parameters for evaluating the quality of many products. While traditional sensory analysis continues to be a valuable method of food and beverage analysis, it is not without its limitations in the evaluation of certain defects. Recently, we found this to be the case when presented with the problem of identification of the off-odor in real samples of liquid whey.

Whey is the leftover liquid when coagulating milk to produce cheese or the released liquid after the fermentation of other dairy products, most often Greek-style yogurts or skyrs. After the coagulation of milk with enzymatic rennet in the production of cheeses, sweet whey is produced, while the use of lactic acid in the production of curd results in acid whey. These two types differ slightly in their composition (Karagul-Yuceer, Drake and Cadwallader, 2003).

While both types mostly consist of water, lactose, proteins, minerals, and fat, acid whey contains more minerals (especially calcium) and less proteins and lactose than sweet whey (Kilara, 2015). Lactose makes up more than 75% of the total solids and is also the main reason why whey is considered as one of the most polluting food streams. On the other hand, the contained proteins and peptides (mainly α-lactalbumin, β-lactoglobulin, serum albumin, and immunoglobulins) are of exceptional biological and functional value and thus offer a wide range of whey utilization (Anand, Khanal and Chenchaiah, 2013; Smithers, 2008).

Dried or concentrated whey and products where some ingredients, especially proteins, are isolated or concentrated (whey protein hydrolysate (WPH), whey protein concentrate (WPC, 34 – 89% protein) and whey protein isolate (WPI, >90% protein)) are widely used as a food ingredient for human consumption (Evans et al., 2009). Liquid whey drinks are gaining popularity, either native, demineralized, or further processed (fermented, carbonated) and flavored in various ways (Francis, 1999).

Liquid whey beverages are very susceptible to microbial spoilage and associated undesirable qualitative deviations because of their rich nutritional content and high water activity. Their shelf life is ensured either by heat treatment (pasteurization or sterilization) or by fermentation and subsequent cold storage (Lo et al., 2016).

Microorganisms generally spoiling whey are the ones that typically spoil milk. Raw and pasteurized milk, exposed to secondary contamination, is most often contaminated by gram-negative bacteria of the genus Pseudomonas, while pasteurized milk is most often spoiled by thermoduric spore-forming microorganisms of the Bacillus and Paenibacillus species. Since whey is easily subjected to lactose fermentation to produce ethanol, acetic, lactic, and propionic acid (which is formed from lactic acid by the bacteria of the genus Propionibacterium), the naturally present lactic acid bacteria – LAB (e.g, Streptococcus, Lactobacillus and Lactococcus) form a variety of aromatic active compounds such as 2,3-butanedione (diacetyl), acetoin, acetaldehyde or acetic acid from pyruvate, an intermediate in lactose fermentation (Lo et al., 2016). When preparing alcoholic whey beverages, mainly yeasts of the genus Kluyveromyces (K. fragilis and K. marxianus) are used. These beverages are characterized by the presence of volatile compounds, including higher alcohols (mainly isoamyl alcohol, isobutanol, and 1-propanol), ethyl esters (mainly ethyl acetate), as well as acids and acetals (Dragone et al., 2009). All of these substances contribute to the natural aroma of whey, but at higher concentrations to an undesirable odor. However, the main contributors to the off-odor of both liquid and dry whey are lipid oxidation products namely aldehydes, ketones, alcohols, and alkanes (Carunchia Whetstine et al., 2003).

Sensory properties have traditionally been described and evaluated via sensory analysis, which can be loosely divided into two groups: discriminant methods and descriptive methods. The purpose of discrimination testing is to indicate whether a tested sample is perceived as being significantly different from a standard one (e.g. Triangle or Duo-Trio test). Descriptive methods, such as the flavor profile method or quantitative descriptive analyses, are more similar to chemical analysis in that they aim to determine the presence or intensity of a particular characteristic (Kilcast, 2010). The problem is that while descriptive methods can characterize a particular off-odor, they are not able to link it to the specific compound, or compounds, responsible for certain occasional defects. That is where instrumental methods come in.

However, instrumental methods are best used in combination with sensory analysis. For example, GC-MS is able to identify the most abundant volatile compounds in a sample but cannot provide clear information on whether the substances are sensorially active. And the most sensitive physical detectors (MSD, ECD, FID) only have detection limits ranging from 1 to 10 pg, whereas human noses can readily detect to 0.05 pg (Muñoz et al., 2010). Gas chromatography with an olfactometric detector (GC-O) combines the high resolution of capillary gas chromatography with the high selectivity and sensitivity of the human nose to detect and identify the compound, or compounds, responsible for an off-odor. The assessors sniff the eluate from the gas chromatograph using a special olfactory port to detect the presence of sensory-active compounds. Recently, solid-phase microextraction with gas chromatography/mass spectrometry coupled with olfactometry (SPME-GC-MS/O) has been used to identify substances in a variety of matrices, including coffee, cheeses, milk powders, orange juice, cashew apple (Anacardium occidentale) juice, yogurt, and even chocolate. (Zellner et al., 2008; Gocmen et al., 2005; Semmelroch and Grosch, 1995; Zepka et al., 2014)

The storage experiment was designed to copy the recommended storage conditions while promoting and accelerating the development of sensory active substances indicative of undesirable changes. For the samples stored at 6 °C for 1 week, there were no significant changes in coliforms, yeast and mold count, dry matter, and pH, as expected, because the declared storage conditions were 4 – 8 °C with a shelf life of 4 days (Figure 1 shows the average values of the 4 measurements). The whey dry matter content did not change significantly (p >0.05) with the storage temperature. At the same time in samples stored at 12 °C and 25 °C, the pH dropped significantly (p <0.001) by 35%. This pH drop was caused by an increased content of organic acids (probably produced by lactic acid or acetic acid bacteria and yeasts) (Campbell et al., 2011; Sattin et al., 2016), which was confirmed by a subsequent chromatographic analysis of the volatile compounds. The number of coliforms and yeasts in the sample stored at 25 °C increased significantly (p <0.05) (coliforms one hundred times, yeasts almost five thousand times). No molds were detected in any samples.

The profile of volatile substances of fresh whey was very poor. Only hexane and octanoic acid had a peak area higher than 10⁴. Other peaks were either not identified or were contaminants from the chromatographic system (siloxanes, higher hydrocarbons, etc.). The limited sensitivity of the MS detector is unfortunately due to a compromise of possible simultaneous application of olfactometry (see batch separation conditions and split between detectors).

After the storage experiment at 25 °C, the profile of volatiles changed drastically both in the number of peaks as well as in their area. A number of ketones (acetone, 2,3-butanedione, 2-butanone, acetoin, 2,3-pentandione, 2-heptanone, 2-nonanone), sulphur compounds (dimethyl sulphide dimethyl disulphide, dimethyl trisulphide, 2,4-dithiapentane), carbonyl compounds (heptane, hexane), an aldehyde (nonanal), alcohols (1-butanol 1-hexanol, 1-octanol) and organic acids (acetic acid, hexanoic acid, and octanoic acid) were found. We have not identified as many aldehydes as Croissant et al. (2009) but a wider range of compounds more similar to Leksrisompong, Miracle, and Drake (2010) and Liaw et al. (2011) findings, including the not commonly found acetoin (a product of LAB metabolism) (Nadal et al., 2009).

Compared to other dairy products, whey is not very rich in sensory active substances (Fox et al., 2016; Qian and Reineccius, 2003). This finding is in an agreement with our measurements of fresh whey at T0 where only 2 sensory active compounds were detected – not identified (NI) compound and octanoic acid (Table 1). Octanoic acid was found to have a very low odor threshold in air (0.86 μg.kg^-1) (Cometto-Muñiz and Abraham, 2010) as opposed to, for example, hexanal (having 0.14 μg.L^-1 therefore 119 μg.kg^-1 in the air at 25 °C (Ömür-Özbek and Dietrich, 2008). This explains why it was detected at T0, along with the fact that hexanal is an oxidation product from linoleic acid, therefore its concentration increases during storage. The other compound that could not be reliably identified by the spectra NIST library (the probability match was less than 50%) had 133 (100), 151 (65), and 13 5 (26) m/z ions as the largest. Detection of such a compound confirms the higher sensitivity of a human nose compared to a mass detector (Muñoz et al., 2010). In total, seven compounds, mostly with an unpleasant odor, were detected by at least two assessors in the sample stored at 25 °C their odor was compared to literature (“The Good Scents Company,” n.d.).

Dimethyl sulphide is usually associated with a cabbage-like odor produced by cooking certain vegetables and cereals, formed along with dimethyl trisulphide by bacterial degradation of sulphur amino acids (Franco-Luesma and Ferreira, 2016; Luo et al., 2018; Nishibori et al., 2014). 2,3-butanedione, or diacetyl, is a natural by-product of the fermentation of lactic acid by the oxidative decarboxylation of α-acetolactate (Hugenholtz et al., 2000). It may also be formed as an intermediate in high-temperature treatment with non-enzymatic Maillard browning and may later be involved in Strecker degradation with other free amino acids (Smit, Smit and Engels, 2005). It is responsible together with acetoin for the characteristic taste of butter (Karagul-Yuceer, Drake and Cadwallader, 2003). Hexanal is a lipid oxidation product and has been proposed as a potential quality marker. Free fatty acids (hexanoic and octanoic) are formed by microbial hydrolysis of fats (Panseri et al., 2011).

Based on the assessor’s results, dimethyl trisulphide produced the highest intensity odor, even though one assessor did not detect it. There are a number of factors that could have caused it, for example, selective anosmia to sulphur compounds (as dimethyl sulphide was not detected either) or the assessor’s fatigue (Brattoli et al., 2013). Dimethyl trisulfide had a peak approximately 20 times smaller than octanoic acid but since the peak area does not correlate with the odor intensity (Högnadóttir and Rouseff, 2003), olfactometry was able to mark it as one of the main compounds responsible for off-odor.

When storing whey samples at elevated temperatures (12 °C and 25 °C), there was a significant increase (p <0.001) in the number of coliform bacteria and yeasts which led to an increased amount of organic acids and alcohols, causing undesirable off-odors. However, the scientific hypothesis (Determination of sensory active substances allows for quick and early identification of microbial degradation and lipid oxidation) is only partially confirmed. Since the slight (by 1 – 2 order) increase in the number of pathogenic and spoilage microorganisms at low storage temperature was not yet reflected in the sufficient production of secondary volatile metabolites (and therefore on the sensory properties of the product) the SPME-GC-MS/O method was not able to detect it.

Thus we conclude that the traditional microbial testing is irreplaceable for a proper setting of storage conditions and shelf life Olfactometers can then play a significant role in detecting the causes of major product odor changes(as we have shown on the sample stored at 25 °C) and therefore spotting specific signs of microbial and chemical degradation.