EVALUATION OF SENSORY AND ANTIOXIDANT PROPERTIES OF COMMERCIAL COFFEE SUBSTITUTES

Sensory pro ﬁ les of commercial coffee substitutes were determined and their possible interdependences with antioxidant characteristics (FRAP, DPPH, ABTS, and CUPRAC), total polyphenol content, and colour were investigated and compared to coffees. Statistically relevant relations were revealed between certain sensory attributes, colour, and antioxidant capacity. Sensory attributes show distinct patterns for coffees, their blends, and substitutes, but no signi ﬁ cant differences between substitutes from different raw materials were found, except for chicory. Although coffees have generally higher antioxidant capacities than their substitutes, these latter, especially chicory-based products, are also valuable antioxidant sources, as only half of them had signi ﬁ cantly lower polyphenol and antioxidant contents when compared to coffee.Principal component analysis was applied to reveal possible differentiation pattern between samples, based on both their sensory and antioxidant attributes.

Coffee is one of the most frequently consumed beverages worldwide, although there are some concerns related to the negative aspects of its excessive consumption, in particular to the addictive and stimulating effect of caffeine (MAJCHER et al., 2013). Coffee substitutes offer a favourable alternative to real coffee brews both from economical and health-care points of view (BAEZA et al., 2017). Instant products are also available on the market, as well as coffee substitute blends containing variable amounts of coffee.
Raw materials used for coffee substitutes are rich in polyphenols, although their nature differs from those detected in coffee. Chicory contains high amounts of polyphenols (400-600 mg gallic acid equivalents/100 g), the dominant compounds being chicoric acid (HEIMLER et al., 2009), syringic and ferulic acids, and catechin (SAHAN et al., 2017). Barley and rye are also considerable sources of antioxidants (BONDIA-PONS et al., 2009). Additionally, the roasting process of coffee substitutes takes places at temperatures (180-200 ºC), where antioxidant melanoidines are formed in Maillard reaction.
Product quality also involves sensory impressions of coffee and coffee substitutes. In consumer product development, just-about-right scales identify the product attributes that should be changed to achieve higher consumer acceptance (GERE et al., 2015a). Merging consumer data with the results of a trained sensory panel, preference maps can help to * To whom correspondence should be addressed. understand the differences and similarities between samples and their attributes (GERE et al., 2014;2015b). Compared to the literature available on the antioxidant capacity and the sensory profi le of coffee, only limited data are available on its substitutes (MAJCHER et al., 2013;BAEZA et al., 2017;GORJANOVIĆ et al., 2017). The target of the present paper is not only to deliver data on polyphenol content, antioxidant capacity, and sensory profi le of surrogate coffees, but also to reveal the possible relations between their sensory and antioxidant attributes.

Samples
A total of 10 representative samples, purchased in supermarkets, were included in the study (Table 1).

Analytical methods
Five replicates were performed for each analytical measurement. Spectrophotometric measurements were performed with a Thermo Helios Alpha UV-VIS spectrophotometer (±0.001 units of absorbance, 1 cm light path). Samples were prepared for analytical measurements according to instructions from packaging (traditional brewing or instant), infusing 3 g coffee substitute/coffee with 100 ml distilled water for 5 min or, for instant products, dissolving 3 g in 100 ml hot distilled water. Cooled brews were decanted by centrifugation at 200 g for 10 min, supernatants were used diluted accordingly.
1.3.1. Total polyphenol content, ferric reducing antioxidant power (FRAP), DPPH, ABTS, and cupric reduction antioxidant capacity (CUPRAC) assays. Total polyphenol content was evaluated following a method adapted from SINGLETON and ROSSI (1965). In FRAP experiments the procedure described by BENZIE and STRAIN (1996) was adapted. DPPH assay was performed according to the procedure by BRAND-WILLIAMS and co-workers (1995), and inhibition rates (I%) were reported. TEAC (ABTS) antioxidant capacity was adapted from RE and co-workers (1999). CUPRAC assay was performed according to APAK and coworkers (2007).
1.3.2. Colorimetry. L*, a*, b* values were analysed by means of a trichromatic refl ection chromameter Konica Minolta CR-410 (Konica Minolta, Osaka, Japan) using a D65 illuminant and CIE: 2° standard observer. The instrument was standardized against a reference white plate (CR-A44) before sample measurements. Coffee samples were spread out at 1 cm depth on a Petri dish, and the L*, a*, b* -values were measured in triplicate, at several points of each sample.

Sensory analysis
Samples were brewed according to the instructions from packaging. Packaging was not shown to the participants, avoiding thus the effect of brand knowledge (SZŐKE et al., 2012). Sugar was added to samples equalling the amount of sugar in the sugar-containing sample Ch-M. Two groups of samples were set: coffee substitutes (6 samples) and coffees and their blends (4 samples). Twelve trained panellists were given a complete list of the pre-defi ned taste and aroma determinants (see Figs 1 and 2). Measurements were done in two replicates. Neutral mineral water was used as taste neutralizer between evaluations . The sensory tests were carried out meeting all criteria of ISO 13299. The intensity values for the reference samples (sample B-R-Ch_org and B-Ch-M-R-C15%) were pre-defi ned in both groups for each attribute on the evaluation scales.   (KÓKAI et al., 2002;VÁRVÖLGYI et al., 2015). Profi le analysis was evaluated by one-way analysis of variance (ANOVA), at signifi cance level of 5%, followed by LSD (least signifi cant differences) post hoc tests. Panel performance was examined to identify panellists with poor performance .

Polyphenol content and antioxidant capacity
In accordance with previously reported data (KOMES et al., 2015;CONTRERAS-CALDERÓN et al., 2016), total polyphenol content (TPC) showed the highest value for robusta coffee, while arabica had a slightly lower value (Table 2). Coffee blends and pure chicory coffee substitute had polyphenol contents not signifi cantly lower than for robusta, while chicory-containing substitutes showed values not signifi cantly lower than arabica. As there is no universally accepted method available for determination of in vitro antioxidant capacity in foods, a range of assays described in the literature (DPPH, FRAP, ABTS, and CUPRAC) and based on different transfer mechanisms were used in order to achieve a complex characterisation of the coffee substitutes investigated. By applying these, generally similar performances were obtained for the samples, these being in good agreement with their polyphenol content. The values measured are in good agreement with those reported for barley coffee (CARVALHO et al., 2014), chicory (GORJANOVIĆ et al., 2017), and coffee (CONTRERAS-CALDERÓN et al., 2016). Since chicory had the overall best performance, the pure barley substitute being on the other end of the spectrum, it can be concluded thatalthough in the same range -the antioxidant potential of these substitutes shows a different distribution compared to the results reported by Croatian researchers (KOMES et al., 2015;GORJANOVIĆ et al., 2017), where barley products reached less disappointing values compared to chicory.

Sensory analysis
Statistical analysis on signifi cant differences (ANOVA analysis) regarding the sensory attributes of the samples was performed separately for the two sets of samples. Sensory profi le of coffee substitutes (Fig. 1) shows that for pure chicory coffee (Ch), aftertaste and bitterness are signifi cantly more intense than for any other sample, burnt fl avour being also more intense than for many other substitutes. Apart from coffee substitute B-R-Ch_org, chicory has shown signifi cantly more intense taste character and burnt fl avour than any other substitute. This dominancy of bitter and roasted notes was also reported by KOMEs and coworkers (2015). Sample B, containing exclusively barley, is dominated in its aroma by a Acta Alimentaria 48, 2019 strong cereal-like note, this being signifi cantly different only from chicory-based samples Ch and Ch-M. Sample R-B-Ch reached the highest scores for coffee-like notes, the lowest values being obtained for samples Ch-M and B-R-Ch_org. The results show that it is not simply the raw materials, but rather their balance and most probably roasting that will infl uence the desired sensory character.
Sensory profi le of coffee-containing samples (Fig. 2) revealed only two properties for which no signifi cant differences were found between samples: transparency, which is low for all coffee-containing samples, and cereal fl avour. Arabica and robusta differ in colour depth, hue, and sourness. The different composition of coffee blends is refl ected in their sensory attributes, especially cereal odour (but not fl avour), bitterness, burnt fl avour (specifi c for chicory) for sample B-R-Ch-C15%, in which substitutes dominate. When samples B-R-Ch-C15% and B-C47% are compared, the higher coffee content of the latter resulted in a more intensive coffee fl avour, but no signifi cant difference regarding coffee odour, burnt and cereal fl avour. On the other hand, coffees and blends show signifi cant differences in bitterness and sourness (this latter being the highest for arabica). Surprisingly, the high cereal content of the blends has only minimal impact on transparency and cereal fl avour. Interestingly, cereal fl avour was found to be in a relatively narrow range for all substitutes, blends, and coffees (45-63%), regardless of composition (except for sample Ch-M, reaching an outstandingly low score), while the same does not apply for cereal odour.

Colour measurement
Results of measurements of colour parameters such as L* (brightness), a* (red-green scale), b* (yellow-blue scale) and calculated parameters C (chroma, colour strength) and h (hue, colour purity) are presented in Table 3. Kruskal-Wallis analysis of brightness values shows signifi cantly lower values for coffee samples, and the blend containing 47% coffee (B-C47%), coffees, and blends are also generally less yellowish than substitutes. Signifi cant differences between coffee-based products and the rest of the samples are refl ected in colour purity (C) and colour strength (h) as well. Colour differences are attributed to the differences in the type and colour of the melanoidines formed in Maillard process on roasting (ECHAVARRÍA et al., 2013), these being due mainly to the higher fructose content of chicory. Antioxidant characteristics correlate strongly with coffee odour and coffee fl avour, while transparency correlates negatively with all antioxidant values. Hue is strongly linked to antioxidant potential, while brightness is inversely related to it. Brightness is also related to transparency for all samples and negatively correlated to hue, coffee fl avour, and sourness. Sensory attributes were also proven to show signifi cant relationships, e.g. burnt fl avour is strongly linked to taste character, aftertaste, bitter and sour tastes. The fact that bitter and sour tastes and burnt fl avour are interlinked refl ects the impact of Maillard reaction products on fi nal aroma not only in case of coffees (as previously described, e.g. BELITZ et al., 2009), but also for substitutes from various botanical sources. On the other hand, statistical evaluation proved the dependence between TPC and bitter taste, raising the possibility of contribution of bitter polyphenols (GONZALO-DIAGO et al., 2014) to the bitter taste of these products. Among other compounds, melanoidins were also identifi ed as major contributors to bitter taste (SUNARHARUM et al., 2014). Similarly, the bitter taste of coffee substitutes is not easily assigned to specifi c compounds. In case of raw chicory, bitter notes are attributed mainly to sesquiterpene lactones, which are almost completely degraded during roasting (LECLERCQ, 1992), thus, the high polyphenol content of roasted chicory may contribute to the bitterness of chicory coffee substitute. For both chicory and cereal substitutes, melanoidines (resulting partly from polyphenols) will contribute to bitter notes in the fi nal taste profi le (CARVALHO et al., 2014).  (Fig. 3). According to principal components PC1 and PC2, covering 73.60% of total variance, coffees are distinguished by the rest of the samples in terms of coffee notes, cereal odour, and antioxidant capacity. Although there are some specifi c sensory notes related to some of the substitutes (e.g. preponderance of burnt, bitter notes for chicory, apparently missing at the chicory-malt product, situated in the opposite quadrant), coffee substitutes apparently do not show composition-related features in their sensory properties. PC1 did not explain the differences between chicory-and coffeecontaining samples (variables responsible for PC1: polyphenols, antioxidant capacity, burnt, bitter and sour notes, aftertaste), however, this variance is well explained by PC2 (variables responsible for PC2: coffee fl avour, coffee odour, hue, colour depth, transparency). Cerealcontaining substitutes are not differentiated based on PC1 and PC2.

Conclusions
Very high polyphenol contents and antioxidant capacities were detected in coffees, coffee blends, and chicory. Half of the coffee substitutes analysed were found to show statistically insignifi cant differences compared to pure coffees in terms of antioxidant potential.
PCA revealed distinct sample patterns for coffees and substitutes. Correlations between antioxidant properties and polyphenol content, as well as sensory, chromametric, and antioxidant properties were disclosed. It was shown that coffee-like notes are directly linked to antioxidant capacity, while transparency and brightness are negatively correlated to the latter. Burnt, sour, and bitter notes are strongly interlinked, this underlining the importance of Maillard reaction products in the formation of fi nal fl avour notes in case of both coffees and their substitutes.
The positive nutritional properties of coffee substitutes highlighted in this paper, i.e. high antioxidant capacity, especially for chicory-based preparations, together with absence of caffeine can qualify these products as valuable sources of cheap and healthy coffee alternatives, contributing to dietary antioxidant intake of possible target groups like children, pregnant and breastfeeding women, as well as elderly people. * Zs. Bodor was supported by the Doctoral School of Food Science (SZIU).