- Email: [email protected]
Benzaldehyde in cherry flavour as a precursor of benzene formation in beverages
Food Chemistry 206 (2016) 74–77
Contents lists available at ScienceDirect
Food Chemistry journal homepage: www.elsevier.com/locate/foodchem
Short communication
Benzaldehyde in cherry flavour as a precursor of benzene formation in beverages Christine Loch a,b, Helmut Reusch a, Ingrid Ruge a, Rolf Godelmann a, Tabea Pflaum a, Thomas Kuballa a, Sandra Schumacher a, Dirk W. Lachenmeier a,⇑ a b
Chemisches und Veterinäruntersuchungsamt (CVUA) Karlsruhe, Weissenburger Strasse 3, D-76187 Karlsruhe, Germany University of Applied Sciences Bingen, Berlinstrasse 109, D-55411 Bingen, Germany
a r t i c l e
i n f o
Article history: Received 2 December 2015 Received in revised form 29 February 2016 Accepted 11 March 2016 Available online 11 March 2016 Chemical compounds: Benzene (PubChem CID: 241) Benzaldehyde (PubChem CID: 240) Benzoic acid (PubChem CID: 243) Ascorbic Acid (PubChem CID: 54670067) Keywords: Benzene Benzaldehyde Food contamination Beverages Cherry
a b s t r a c t During sampling and analysis of alcohol-free beverages for food control purposes, a comparably high contamination of benzene (up to 4.6 lg/L) has been detected in cherry-flavoured products, even when they were not preserved using benzoic acid (which is a known precursor of benzene formation). There has been some speculation in the literature that formation may occur from benzaldehyde, which is contained in natural and artificial cherry flavours. In this study, model experiments were able to confirm that benzaldehyde does indeed degrade to benzene under heating conditions, and especially in the presence of ascorbic acid. Analysis of a large collective of authentic beverages from the market (n = 170) further confirmed that benzene content is significantly correlated to the presence of benzaldehyde (r = 0.61, p < 0.0001). In the case of cherry flavoured beverages, industrial best practices should include monitoring for benzene. Formulations containing either benzoic acid or benzaldehyde in combination with ascorbic acid should be avoided. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Benzene is one of the compounds with the highest level of evidence that it may cause cancer in humans (Steinbrenner, LöbellBehrends, Reusch, Kuballa, & Lachenmeier, 2010). For example, it was classified into group 1 by the International Agency for Research on Cancer (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2012). For this reason, levels of benzene in foods ‘‘shall be kept as low as can reasonably be achieved by following good practices” (ALARA principle) according to the EU regulation on food contaminants (European Council, 1993), and for some food groups such as drinking water maximum limits have been implemented by law; for example a maximum limit of 1.0 lg/L is demanded by the EU drinking water directive ⇑ Corresponding author. E-mail addresses: [email protected] (C. Loch), [email protected]. de (H. Reusch), [email protected] (I. Ruge), [email protected]. de (R. Godelmann), [email protected] (T. Pflaum), Thomas.kuballa@ cvuaka.bwl.de (T. Kuballa), [email protected] (S. Schumacher), [email protected] (D.W. Lachenmeier). http://dx.doi.org/10.1016/j.foodchem.2016.03.034 0308-8146/Ó 2016 Elsevier Ltd. All rights reserved.
(European Council, 1998). There may be two different mechanisms causing the occurrence of benzene in foods and beverages: the products may either be (often inadvertently) contaminated from external sources containing benzene (such as petrol), or benzene may be formed intrinsically as a heat-induced contaminant from various natural or artificial precursors found in the foods (Lachenmeier, Kuballa, et al., 2010; Lachenmeier, Steinbrenner, Löbell-Behrends, Reusch, & Kuballa, 2010; Salviano dos Santos, Medeiros Salgado, Guedes Torres, & Signori Pereira, 2015). While the first mechanism is well manageable and has not led to large problems except for isolated cases, e.g., due to contaminated carbon dioxide used in beverage production (Wu, Lin, Fan, Dong, & Chen, 2006), the second mechanism was of considerable concern when benzene contamination was regularly detected in alcoholfree beverages with levels of up to 23 lg/L (Cao, Casey, Seaman, Tague, & Becalski, 2007; Gardner & Lawrence, 1993; Lachenmeier, Reusch, Sproll, Schoeberl, & Kuballa, 2008; McNeal, Nyman, Diachenko, & Hollifield, 1993; Nyman et al., 2008; Page, Conacher, Weber, & Lacroix, 1992). The source of this contamination, benzoic acid used as an additive for preservation, was quickly
C. Loch et al. / Food Chemistry 206 (2016) 74–77
detected. Benzoic acid may deteriorate to considerable contents of benzene, especially when used in combination with ascorbic acid (Salviano dos Santos et al., 2015). Industry guidelines were established suggesting the replacement of benzoic acid with other preservatives, specifically when used in combination with ascorbic acid (either naturally contained or used as an antioxidant additive) (Salviano dos Santos et al., 2015). It was therefore surprising when consumer magazines reported the occurrence of comparably high contents of benzene in beverages not preserved with benzoic acid. A concentration of 4.6 lg/L was detected in a cherry beverage (Stiftung Warentest, 2013b), which led to a larger survey of cherry beverages (n = 30), from which 6 were found positive, with levels between 0.3 and 2.5 lg/L of benzene. As the benzene levels were roughly correlated to the occurrence of benzaldehyde (9.5–135 mg/L), this led to the hypothesis that benzaldehyde may be the precursor of benzene formation (Stiftung Warentest, 2013a). The scientific literature offers only limited evidence to corroborate this hypothesis. McNeal et al. (1993) showed in a model solution that 74 lg/kg benzene may form from benzaldehyde at 0.04% (w/w) in water in the presence of 0.025% ascorbic acid. Nyman, Wamer, Begley, Diachenko, and Perfetti (2010) detected the formation of 0.3 lg/kg benzene in a cherry flavoured drink after storage for 14 days at 40 °C. The benzaldehyde content of the drink was 54 mg/kg. This study was conducted to provide evidence into the potential mechanism of benzene formation in beverages from benzaldehyde by model experiments and correlation analysis of data from a large market survey. 2. Materials and methods 2.1. Survey of cherry-flavoured beverages Between January and March 2014, 170 samples submitted to the CVUA Karlsruhe were analysed for benzene, benzaldehyde and benzoic acid. As part of official food control, our institute is the central beverage control laboratory in Baden-Württemberg, a German federal state with a population of approximately 10.6 million. The sampling has been done by local authorities directly at food producers or retail trade. The samples have been randomly selected and collected by government food inspectors, as our institute has requested only the food group and sample number to collect but not specific brands. For this survey, we requested the food groups ‘‘soft drinks with cherry flavour”, and ‘‘mineral or table water flavoured with cherry”. Some additional samples (about 10%) were directly purchased over the internet or in wholesale stores in Karlsruhe, Germany. In total, we received 72 cherry-flavoured samples. Additionally, 98 soft drinks and flavoured water samples with other flavours except cherry were analysed for comparison. Non-parametric statistical comparisons (Mann–Whitney tests) between groups of samples were conducted, using the software package Stat Tools for Excel Version 5.5.0 (Palisade Corporation, Ithaca, NY). Linear correlation analysis was conducted using Origin V.7.5 (Originlab, Northampton, MA). 2.2. Benzene formation from pure benzaldehyde (model experiment) In order to evaluate the possible benzene formation capability of benzaldehyde in aqueous solution with a minimum amount of experiments, a D-optimal experimental design was used. The D-optimal algorithm chooses an ideal subset of all possible combinations and significantly reduces the number of required experiments compared to standard design types. To simulate the matrix of cherry-flavoured beverages, a citrate buffer at pH 3.5 was chosen as the medium (the pH was chosen because
75
alcohol-free drinks typically have pH values between 3 and 4). Benzaldehyde and/or ascorbic acid were added at a concentration of 75 mg/L or 90 mg/L, respectively. The heating temperature was studied at three levels (unheated i.e. room temperature at about 20 °C, 60 °C, and 100 °C). In the case of the heated samples, the heating time was varied at two levels (2 h, and 24 h). The experiments were conducted with 10 mL of liquid directly in the headspace vials that were measured afterwards without opening to avoid any loss of benzene. In total, 23 experiments were conducted in duplicate (n = 46). The experimental designs and calculations were done using the Software Package Design Expert V7.0.0 (Stat-Ease Inc., Minneapolis, MN). 2.3. Chemical analysis Benzene was analysed using a validated headspace-gas chromatography/mass spectrometry (HS-GC/MS) procedure with a deuterated internal standard, previously described in detail (Lachenmeier et al., 2010; Lachenmeier et al., 2008). The quantification was conducted in the selected ion monitoring (SIM) mode; for benzene: m/z 78 as target ion and m/z 77 as qualifier ion, and for benzene-d6 as internal standard: m/z 84 as target ion and m/z 82 as qualifier ion. Determined according to the German norm DIN 32645, the limit of detection was 0.03 lg/L and the limit of quantitation was 0.09 lg/L. Benzaldehyde was analysed using a separate HS-GC/MS procedure. For this, 1 mL of sample was mixed with 4 mL of water, 100 lL of internal standard (benzaldehyde-d6, 40 lg/mL in methanol) and 100 lL of methanol. Using 100 lL of 25% KOH, the solution was adjusted to approximately pH 10 to neutralise the carbonic acid contained in some samples. The quantification was conducted in the selected ion monitoring (SIM) mode; for benzaldehyde: m/z 106 as target ion and m/z 105 and m/z 77 as qualifier ions, and for benzaldehyde-d6 as internal standard: m/z 112 as target ion and m/z 110 as qualifier ion. The limit of detection was 0.5 mg/L and the limit of quantitation was 1.7 mg/L. Finally, benzoic acid was determined using HPLC according to the German reference method (Anon, 1984). The limit of detection was 0.5 mg/L and the limit of quantitation was 1.5 mg/L. 3. Results The results of the model experiment are shown in Fig. 1. The analysis of variance (ANOVA) for the response surface quadratic model implied that the model is significant (p < 0.0001, F-value 23.18) with a coefficient of determination (r2) of 0.9040. The model parameters were judged as adequate to navigate the design space. The following model terms were significant: heating time (p = 0.0002), benzaldehyde (p < 0.0001), ascorbic acid (p < 0.0001) and temperature (p < 0.0001). The following interactions between model terms were significant: heating time-benzaldehyde (p = 0.0024), heating time-ascorbic acid (p = 0.0027), benzaldehyde-ascorbic acid (p < 0.0001), benzaldehydetemperature (p < 0.0001), ascorbic acid-temperature (p = 0.0002). It is therefore obvious that benzene may be formed from benzaldehyde under heating, and to a greater extent in the presence of ascorbic acid. The results of the survey of benzene, benzaldehyde and benzoic acid in 170 beverages are shown in Table 1. The average benzene concentration in cherry-flavoured beverages was 0.18 lg/L, while the value in beverages with other flavours was 0.08 lg/L. A significant linear relation between benzaldehyde and benzene concentrations was detected (r = 0.6122, p < 0.0001) (Fig. 2), but not between benzoic acid and benzene concentrations (r = 0.1016, p = 0.2338). When the samples were sorted in benzene negative
76
C. Loch et al. / Food Chemistry 206 (2016) 74–77
24
Benzene (µg/L)
Heating time = 24 h Benzaldehyde = present 18
12
6
0
present
100
60
Temperature (°C)
20
Ascorbic acid
not present
Fig. 1. Heat-induced formation of benzene from benzaldehyde in model experiments. Significant amounts of benzene are formed if the beverage is heated at temperatures higher than 60 °C especially in the presence of ascorbic acid.
and benzene positive samples (Table 2), the benzaldehyde concentration was significantly higher in the benzene positive samples, while for benzoic acid no significant difference between the two groups was observed.
4. Discussion Benzaldehyde is used as an artificial essential oil of almond (Burdock, 2009) and typically contained in cherry flavours. According to data from the Flavor and Extract Manufacturers Association (FEMA), benzaldehyde is reportedly used in non-alcoholic beverages in usual concentrations of 29 mg/L and a maximum of 58 mg/L (Burdock, 2009). We cannot confirm these reports for Germany, as our beverages contained much lower concentrations of benzaldehyde with an average of 8 mg/L and a maximum of 42 mg/L. If a higher benzaldehyde dosage might still occur in the USA, there might be an even higher potential for benzene formation than in our samples. We believe that our research provides strong evidence for the causal relationship between benzaldehyde and benzene formation in beverages. Our model experiment proved that benzene can indeed be formed from benzaldehyde in aqueous solutions with beverage-typical slightly acidic pH and under heating conditions
(which may occur for example in the sterilisation step of beverage filling). Our large survey provided additional evidence that a relationship between benzene and benzaldehyde exists, while benzoic acid could be excluded as a confounding factor. It should be noted that the correlation (r = 0.6122) between benzene and benzaldehyde in our survey is higher than what has previously been reported in similar evaluations, e.g. between benzene and benzoic acid (r = 0.1827, p = 0.2790) or between benzene and iron (r = 0.2583, p = 0.0098) (Lachenmeier et al., 2008). From a mechanistic side, it has been known from model experiments under quite strong conditions (heating at 250 °C for 5 days) that aquathermolysis of benzaldehyde to benzoic acid may occur (Katritzky, Balasubramanian, & Siskin, 1990). As detailed in the introduction, benzoic acid is a well-known precursor of benzene in beverages (Salviano dos Santos et al., 2015). Becalski and Nyman (2009) judged the direct oxidation of benzaldehyde or the Cannizzaro reaction as unlikely, and suggested that the mechanism may be similar to the decomposition of benzoic acid, i.e. a loss of carbon monoxide from a benzyl radical. On the other hand, the lack of correlation between benzene and benzoic acid in our experiments puts the hypothesis of benzoic acid as intermediary in the formation at least into question, while a much more sensitive method for determination of benzoic acid might be required to detect lg/L-levels that could react to give benzene. Further research into the clarification of the mechanism and possibilities for its inhibition appears to be necessary. Further mechanistic research should also include the influence of transition metals, such as copper and iron, which were suggested as catalysts in benzene formation in beverages in the presence of benzoic acid sources and ascorbic acid (International Council of Beverages Associations, 2006; Lachenmeier et al., 2008). The awareness about the benzene problem in the drinks industry may have increased over the last years and the efforts may have led to reduced concentrations. For example, in our sample from 2006 to 2007 (n = 313) we had a higher incidence of samples exceeding the water limit (n = 7, 2%), a higher average (0.42 lg/L) and some extreme observations up to 42 lg/L (Lachenmeier et al., 2008). If the cherry-flavoured beverages from the current survey are excluded, the benzene average for this study was significantly lower than the one from 2006 to 2007 (p < 0.0001). If the cherry-flavoured beverages are included, no significant difference was detectable, however (p = 0.3350).
5. Conclusions In the case of cherry-flavoured beverages or to be more specific in the case of benzaldehyde-containing beverages, industrial best practices should include monitoring for benzene. The International Council of Beverages Associations (2006) guidance
Table 1 Benzene, benzaldehyde, and benzoic acid concentrations of 170 beverages. Cherry-flavoured beverages (n = 72)
Positive >EU benzene water limit (1 lg/l) Average* Standard deviation* Minimum* P90* P95* Maximum* Median* *
Beverages with other flavours (n = 98)
Benzene [lg/L]
Benzaldehyde [mg/L]
Benzoic acid [mg/L]
Benzene [lg/L]
Benzaldehyde [mg/L]
Benzoic acid [mg/L]
54 (75%) 1 (1%) 0.18 0.35 n.d. 0.39 0.53 2.86 0.11
64 (89%) – 8 9 n.d. 20 29 42 5
19 (26%) – 5 21 n.d. 0 20 123 n.d.
19 (19%) 1 (1%) 0.08 0.55 n.d. 0.07 0.09 5.43 n.d.
9 (9%) – 0.1 0.4 n.d. 0.0 0.6 3.8 n.d.
30 (31%) – 18 43 n.d. 114 129 134 n.d.
Distribution calculated with not detectable (n.d.) results below LOD set as zero and values between LOD and LOQ set at the actual numerical value.
C. Loch et al. / Food Chemistry 206 (2016) 74–77
77
Acknowledgments
0.7
The skilful technical assistance of H. Havel and H. Mann is gratefully acknowledged.
0.6 0.5
References 0.4 0.3 0.2 0.1
Linear Fit 95% Prediction Limit R=0.6122 p<0.0001
0.0
0
10
20
30
40
50
Benzaldehyde (mg/L) Fig. 2. Relationship between benzaldehyde and benzene concentrations in a survey of 170 beverages (results below LOD were calculated as zero, two outliers with extreme benzene concentrations were removed).
Table 2 Benzaldehyde and benzoic acid concentrations of benzene-positive and benzene negative samples in a survey of 170 beverages.
Benzene negative samples Benzene positive samples p (Mann–Whitney Test)
Benzaldehyde [mg/L]
Benzoic acid [mg/L]
0.5 ± 1.6 7.5 ± 9.1 <0.0001
11 ± 34 14 ± 38 0.3730
document to mitigate the potential for benzene formation in beverages, for example, has included the check for benzaldehyde as a control point (CP) that beverage developers may wish to consider when formulating a product (similar to the CP for benzoic acid). Formulations containing either benzoic acid or benzaldehyde in combination with ascorbic acid should be avoided. This research suggests that while the awareness about benzoic acid as a CP is high (as documented by the falling levels over the past decade), this may be less the case about benzaldehyde. The industry should implement mitigation measures for benzene in cherry-flavoured beverages in consideration of the ALARA principle. For the beverage and the flavour companies supplying cherry flavourings this demand may present certain problems. The flavouring substance benzaldehyde is an important component of cherry flavourings. However, it may be possible to substitute it with other flavouring substances with cherry/almond notes. This may provide an alternative solution to the benzene problem by eliminating benzaldehyde.
Conflict of interest statement The authors declare that there are no conflicts of interest.
Anon (1984). Bestimmung von Konservierungsstoffen in fettarmen Lebensmitteln. Amtliche Sammlung von Untersuchungsverfahren, L00.00-9, 1–4 (in German). Becalski, A., & Nyman, P. (2009). Benzene. In R. H. Stadler & D. R. Lineback (Eds.), Process-induced food toxicants: occurrence, formation, mitigation, and health risks (pp. 413–444). Hoboken, NJ, USA: John Wiley & Sons Inc. Burdock, G. A. (2009). Fenaroli’s handbook of flavour ingredients (6th ed.) Boca Raton, FL: CRC Press. Cao, X. L., Casey, V., Seaman, S., Tague, B., & Becalski, A. (2007). Determination of benzene in soft drinks and other beverages by isotope dilution headspace gas chromatography/mass spectrometry. Journal of AOAC International, 90, 479–484. European Council (1993). Council Regulation (EEC) No 315/93 laying down Community procedures for contaminants in food. Official Journal of the European Communities, L37, 1–3. European Council (1998). Council Directive 98/83/EC on the quality of water intended for human consumption. Official Journal of the European Communities, L330, 32–54. Gardner, L. K., & Lawrence, G. D. (1993). Benzene production from decarboxylation of benzoic acid in the presence of ascorbic acid and a transition-metal catalyst. Journal of Agricultural and Food Chemistry, 41, 693–695. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans (2012). Benzene. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, 100F, 249–294. International Council of Beverages Associations (2006). ICBA Guidance Document to mitigate the potential for benzene formation in beverages. Brussels, Belgium: UNESDA – Union of European Beverages Associations. Katritzky, A. R., Balasubramanian, M., & Siskin, M. (1990). Aqueous hightemperature chemistry of carbo-and heterocycles. 2. Monosubstituted benzenes: benzyl alcohol, benzaldehyde and benzoic acid. Energy & Fuels, 4, 499–505. Lachenmeier, D. W., Kuballa, T., Reusch, H., Sproll, C., Kersting, M., & Alexy, U. (2010). Benzene in infant carrot juice: further insight into formation mechanism and risk assessment including consumption data from the DONALD study. Food and Chemical Toxicology, 48, 291–297. Lachenmeier, D. W., Reusch, H., Sproll, C., Schoeberl, K., & Kuballa, T. (2008). Occurence of benzene as heat-induced contaminant of carrot juice for babies in a general survey of beverages. Food Additives and Contaminants, 25, 1216–1224. Lachenmeier, D. W., Steinbrenner, N., Löbell-Behrends, S., Reusch, H., & Kuballa, T. (2010). Benzene contamination in heat-treated carrot products including baby foods. The Open Toxicology Journal, 4, 39–42. McNeal, T. P., Nyman, P. J., Diachenko, G. W., & Hollifield, H. C. (1993). Survey of benzene in foods by using headspace concentration techniques and capillary gas-chromatography. Journal of AOAC International, 76, 1213–1219. Nyman, P. J., Diachenko, G. W., Perfetti, G. A., McNeal, T. P., Hiatt, M. H., & Morehouse, K. M. (2008). Survey results of benzene in soft drinks and other beverages by headspace gas chromatography/mass spectrometry. Journal of Agricultural and Food Chemistry, 56, 571–576. Nyman, P. J., Wamer, W. G., Begley, T. H., Diachenko, G. W., & Perfetti, G. A. (2010). Evaluation of accelerated UV and thermal testing for benzene formation in beverages containing benzoate and ascorbic acid. Journal of Food Science, 75, C263–C267. Page, B. D., Conacher, H. B. S., Weber, D., & Lacroix, G. (1992). A survey of benzene in fruits and retail fruit juices, fruit drinks, and soft drinks. Journal of AOAC International, 75, 334–340. Salviano dos Santos, V. P., Medeiros Salgado, A., Guedes Torres, A, & Signori Pereira, K. (2015). Benzene as a chemical hazard in processed foods. International Journal of Food Science Article ID 545640. Steinbrenner, N., Löbell-Behrends, S., Reusch, H., Kuballa, T., & Lachenmeier, D. W. (2010). Benzol in Lebensmitteln – ein Überblick. Journal für Verbraucherschutz und Lebensmittelsicherheit, 5, 443–452 (in German). Stiftung Warentest (2013a). Kritisches Kirscharoma. Test, 7, 30–33 (in German). Stiftung Warentest (2013b). Kunstaroma statt Frucht. Test, 5, 20–26 (in German). Wu, Q. J., Lin, H., Fan, W., Dong, J. J., & Chen, H. L. (2006). Investigation into benzene, trihalomethanes and formaldehyde in Chinese lager beers. Journal of the Institute of Brewing, 112, 291–294.
Contents lists available at ScienceDirect
Food Chemistry journal homepage: www.elsevier.com/locate/foodchem
Short communication
Benzaldehyde in cherry flavour as a precursor of benzene formation in beverages Christine Loch a,b, Helmut Reusch a, Ingrid Ruge a, Rolf Godelmann a, Tabea Pflaum a, Thomas Kuballa a, Sandra Schumacher a, Dirk W. Lachenmeier a,⇑ a b
Chemisches und Veterinäruntersuchungsamt (CVUA) Karlsruhe, Weissenburger Strasse 3, D-76187 Karlsruhe, Germany University of Applied Sciences Bingen, Berlinstrasse 109, D-55411 Bingen, Germany
a r t i c l e
i n f o
Article history: Received 2 December 2015 Received in revised form 29 February 2016 Accepted 11 March 2016 Available online 11 March 2016 Chemical compounds: Benzene (PubChem CID: 241) Benzaldehyde (PubChem CID: 240) Benzoic acid (PubChem CID: 243) Ascorbic Acid (PubChem CID: 54670067) Keywords: Benzene Benzaldehyde Food contamination Beverages Cherry
a b s t r a c t During sampling and analysis of alcohol-free beverages for food control purposes, a comparably high contamination of benzene (up to 4.6 lg/L) has been detected in cherry-flavoured products, even when they were not preserved using benzoic acid (which is a known precursor of benzene formation). There has been some speculation in the literature that formation may occur from benzaldehyde, which is contained in natural and artificial cherry flavours. In this study, model experiments were able to confirm that benzaldehyde does indeed degrade to benzene under heating conditions, and especially in the presence of ascorbic acid. Analysis of a large collective of authentic beverages from the market (n = 170) further confirmed that benzene content is significantly correlated to the presence of benzaldehyde (r = 0.61, p < 0.0001). In the case of cherry flavoured beverages, industrial best practices should include monitoring for benzene. Formulations containing either benzoic acid or benzaldehyde in combination with ascorbic acid should be avoided. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Benzene is one of the compounds with the highest level of evidence that it may cause cancer in humans (Steinbrenner, LöbellBehrends, Reusch, Kuballa, & Lachenmeier, 2010). For example, it was classified into group 1 by the International Agency for Research on Cancer (IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, 2012). For this reason, levels of benzene in foods ‘‘shall be kept as low as can reasonably be achieved by following good practices” (ALARA principle) according to the EU regulation on food contaminants (European Council, 1993), and for some food groups such as drinking water maximum limits have been implemented by law; for example a maximum limit of 1.0 lg/L is demanded by the EU drinking water directive ⇑ Corresponding author. E-mail addresses: [email protected] (C. Loch), [email protected]. de (H. Reusch), [email protected] (I. Ruge), [email protected]. de (R. Godelmann), [email protected] (T. Pflaum), Thomas.kuballa@ cvuaka.bwl.de (T. Kuballa), [email protected] (S. Schumacher), [email protected] (D.W. Lachenmeier). http://dx.doi.org/10.1016/j.foodchem.2016.03.034 0308-8146/Ó 2016 Elsevier Ltd. All rights reserved.
(European Council, 1998). There may be two different mechanisms causing the occurrence of benzene in foods and beverages: the products may either be (often inadvertently) contaminated from external sources containing benzene (such as petrol), or benzene may be formed intrinsically as a heat-induced contaminant from various natural or artificial precursors found in the foods (Lachenmeier, Kuballa, et al., 2010; Lachenmeier, Steinbrenner, Löbell-Behrends, Reusch, & Kuballa, 2010; Salviano dos Santos, Medeiros Salgado, Guedes Torres, & Signori Pereira, 2015). While the first mechanism is well manageable and has not led to large problems except for isolated cases, e.g., due to contaminated carbon dioxide used in beverage production (Wu, Lin, Fan, Dong, & Chen, 2006), the second mechanism was of considerable concern when benzene contamination was regularly detected in alcoholfree beverages with levels of up to 23 lg/L (Cao, Casey, Seaman, Tague, & Becalski, 2007; Gardner & Lawrence, 1993; Lachenmeier, Reusch, Sproll, Schoeberl, & Kuballa, 2008; McNeal, Nyman, Diachenko, & Hollifield, 1993; Nyman et al., 2008; Page, Conacher, Weber, & Lacroix, 1992). The source of this contamination, benzoic acid used as an additive for preservation, was quickly
C. Loch et al. / Food Chemistry 206 (2016) 74–77
detected. Benzoic acid may deteriorate to considerable contents of benzene, especially when used in combination with ascorbic acid (Salviano dos Santos et al., 2015). Industry guidelines were established suggesting the replacement of benzoic acid with other preservatives, specifically when used in combination with ascorbic acid (either naturally contained or used as an antioxidant additive) (Salviano dos Santos et al., 2015). It was therefore surprising when consumer magazines reported the occurrence of comparably high contents of benzene in beverages not preserved with benzoic acid. A concentration of 4.6 lg/L was detected in a cherry beverage (Stiftung Warentest, 2013b), which led to a larger survey of cherry beverages (n = 30), from which 6 were found positive, with levels between 0.3 and 2.5 lg/L of benzene. As the benzene levels were roughly correlated to the occurrence of benzaldehyde (9.5–135 mg/L), this led to the hypothesis that benzaldehyde may be the precursor of benzene formation (Stiftung Warentest, 2013a). The scientific literature offers only limited evidence to corroborate this hypothesis. McNeal et al. (1993) showed in a model solution that 74 lg/kg benzene may form from benzaldehyde at 0.04% (w/w) in water in the presence of 0.025% ascorbic acid. Nyman, Wamer, Begley, Diachenko, and Perfetti (2010) detected the formation of 0.3 lg/kg benzene in a cherry flavoured drink after storage for 14 days at 40 °C. The benzaldehyde content of the drink was 54 mg/kg. This study was conducted to provide evidence into the potential mechanism of benzene formation in beverages from benzaldehyde by model experiments and correlation analysis of data from a large market survey. 2. Materials and methods 2.1. Survey of cherry-flavoured beverages Between January and March 2014, 170 samples submitted to the CVUA Karlsruhe were analysed for benzene, benzaldehyde and benzoic acid. As part of official food control, our institute is the central beverage control laboratory in Baden-Württemberg, a German federal state with a population of approximately 10.6 million. The sampling has been done by local authorities directly at food producers or retail trade. The samples have been randomly selected and collected by government food inspectors, as our institute has requested only the food group and sample number to collect but not specific brands. For this survey, we requested the food groups ‘‘soft drinks with cherry flavour”, and ‘‘mineral or table water flavoured with cherry”. Some additional samples (about 10%) were directly purchased over the internet or in wholesale stores in Karlsruhe, Germany. In total, we received 72 cherry-flavoured samples. Additionally, 98 soft drinks and flavoured water samples with other flavours except cherry were analysed for comparison. Non-parametric statistical comparisons (Mann–Whitney tests) between groups of samples were conducted, using the software package Stat Tools for Excel Version 5.5.0 (Palisade Corporation, Ithaca, NY). Linear correlation analysis was conducted using Origin V.7.5 (Originlab, Northampton, MA). 2.2. Benzene formation from pure benzaldehyde (model experiment) In order to evaluate the possible benzene formation capability of benzaldehyde in aqueous solution with a minimum amount of experiments, a D-optimal experimental design was used. The D-optimal algorithm chooses an ideal subset of all possible combinations and significantly reduces the number of required experiments compared to standard design types. To simulate the matrix of cherry-flavoured beverages, a citrate buffer at pH 3.5 was chosen as the medium (the pH was chosen because
75
alcohol-free drinks typically have pH values between 3 and 4). Benzaldehyde and/or ascorbic acid were added at a concentration of 75 mg/L or 90 mg/L, respectively. The heating temperature was studied at three levels (unheated i.e. room temperature at about 20 °C, 60 °C, and 100 °C). In the case of the heated samples, the heating time was varied at two levels (2 h, and 24 h). The experiments were conducted with 10 mL of liquid directly in the headspace vials that were measured afterwards without opening to avoid any loss of benzene. In total, 23 experiments were conducted in duplicate (n = 46). The experimental designs and calculations were done using the Software Package Design Expert V7.0.0 (Stat-Ease Inc., Minneapolis, MN). 2.3. Chemical analysis Benzene was analysed using a validated headspace-gas chromatography/mass spectrometry (HS-GC/MS) procedure with a deuterated internal standard, previously described in detail (Lachenmeier et al., 2010; Lachenmeier et al., 2008). The quantification was conducted in the selected ion monitoring (SIM) mode; for benzene: m/z 78 as target ion and m/z 77 as qualifier ion, and for benzene-d6 as internal standard: m/z 84 as target ion and m/z 82 as qualifier ion. Determined according to the German norm DIN 32645, the limit of detection was 0.03 lg/L and the limit of quantitation was 0.09 lg/L. Benzaldehyde was analysed using a separate HS-GC/MS procedure. For this, 1 mL of sample was mixed with 4 mL of water, 100 lL of internal standard (benzaldehyde-d6, 40 lg/mL in methanol) and 100 lL of methanol. Using 100 lL of 25% KOH, the solution was adjusted to approximately pH 10 to neutralise the carbonic acid contained in some samples. The quantification was conducted in the selected ion monitoring (SIM) mode; for benzaldehyde: m/z 106 as target ion and m/z 105 and m/z 77 as qualifier ions, and for benzaldehyde-d6 as internal standard: m/z 112 as target ion and m/z 110 as qualifier ion. The limit of detection was 0.5 mg/L and the limit of quantitation was 1.7 mg/L. Finally, benzoic acid was determined using HPLC according to the German reference method (Anon, 1984). The limit of detection was 0.5 mg/L and the limit of quantitation was 1.5 mg/L. 3. Results The results of the model experiment are shown in Fig. 1. The analysis of variance (ANOVA) for the response surface quadratic model implied that the model is significant (p < 0.0001, F-value 23.18) with a coefficient of determination (r2) of 0.9040. The model parameters were judged as adequate to navigate the design space. The following model terms were significant: heating time (p = 0.0002), benzaldehyde (p < 0.0001), ascorbic acid (p < 0.0001) and temperature (p < 0.0001). The following interactions between model terms were significant: heating time-benzaldehyde (p = 0.0024), heating time-ascorbic acid (p = 0.0027), benzaldehyde-ascorbic acid (p < 0.0001), benzaldehydetemperature (p < 0.0001), ascorbic acid-temperature (p = 0.0002). It is therefore obvious that benzene may be formed from benzaldehyde under heating, and to a greater extent in the presence of ascorbic acid. The results of the survey of benzene, benzaldehyde and benzoic acid in 170 beverages are shown in Table 1. The average benzene concentration in cherry-flavoured beverages was 0.18 lg/L, while the value in beverages with other flavours was 0.08 lg/L. A significant linear relation between benzaldehyde and benzene concentrations was detected (r = 0.6122, p < 0.0001) (Fig. 2), but not between benzoic acid and benzene concentrations (r = 0.1016, p = 0.2338). When the samples were sorted in benzene negative
76
C. Loch et al. / Food Chemistry 206 (2016) 74–77
24
Benzene (µg/L)
Heating time = 24 h Benzaldehyde = present 18
12
6
0
present
100
60
Temperature (°C)
20
Ascorbic acid
not present
Fig. 1. Heat-induced formation of benzene from benzaldehyde in model experiments. Significant amounts of benzene are formed if the beverage is heated at temperatures higher than 60 °C especially in the presence of ascorbic acid.
and benzene positive samples (Table 2), the benzaldehyde concentration was significantly higher in the benzene positive samples, while for benzoic acid no significant difference between the two groups was observed.
4. Discussion Benzaldehyde is used as an artificial essential oil of almond (Burdock, 2009) and typically contained in cherry flavours. According to data from the Flavor and Extract Manufacturers Association (FEMA), benzaldehyde is reportedly used in non-alcoholic beverages in usual concentrations of 29 mg/L and a maximum of 58 mg/L (Burdock, 2009). We cannot confirm these reports for Germany, as our beverages contained much lower concentrations of benzaldehyde with an average of 8 mg/L and a maximum of 42 mg/L. If a higher benzaldehyde dosage might still occur in the USA, there might be an even higher potential for benzene formation than in our samples. We believe that our research provides strong evidence for the causal relationship between benzaldehyde and benzene formation in beverages. Our model experiment proved that benzene can indeed be formed from benzaldehyde in aqueous solutions with beverage-typical slightly acidic pH and under heating conditions
(which may occur for example in the sterilisation step of beverage filling). Our large survey provided additional evidence that a relationship between benzene and benzaldehyde exists, while benzoic acid could be excluded as a confounding factor. It should be noted that the correlation (r = 0.6122) between benzene and benzaldehyde in our survey is higher than what has previously been reported in similar evaluations, e.g. between benzene and benzoic acid (r = 0.1827, p = 0.2790) or between benzene and iron (r = 0.2583, p = 0.0098) (Lachenmeier et al., 2008). From a mechanistic side, it has been known from model experiments under quite strong conditions (heating at 250 °C for 5 days) that aquathermolysis of benzaldehyde to benzoic acid may occur (Katritzky, Balasubramanian, & Siskin, 1990). As detailed in the introduction, benzoic acid is a well-known precursor of benzene in beverages (Salviano dos Santos et al., 2015). Becalski and Nyman (2009) judged the direct oxidation of benzaldehyde or the Cannizzaro reaction as unlikely, and suggested that the mechanism may be similar to the decomposition of benzoic acid, i.e. a loss of carbon monoxide from a benzyl radical. On the other hand, the lack of correlation between benzene and benzoic acid in our experiments puts the hypothesis of benzoic acid as intermediary in the formation at least into question, while a much more sensitive method for determination of benzoic acid might be required to detect lg/L-levels that could react to give benzene. Further research into the clarification of the mechanism and possibilities for its inhibition appears to be necessary. Further mechanistic research should also include the influence of transition metals, such as copper and iron, which were suggested as catalysts in benzene formation in beverages in the presence of benzoic acid sources and ascorbic acid (International Council of Beverages Associations, 2006; Lachenmeier et al., 2008). The awareness about the benzene problem in the drinks industry may have increased over the last years and the efforts may have led to reduced concentrations. For example, in our sample from 2006 to 2007 (n = 313) we had a higher incidence of samples exceeding the water limit (n = 7, 2%), a higher average (0.42 lg/L) and some extreme observations up to 42 lg/L (Lachenmeier et al., 2008). If the cherry-flavoured beverages from the current survey are excluded, the benzene average for this study was significantly lower than the one from 2006 to 2007 (p < 0.0001). If the cherry-flavoured beverages are included, no significant difference was detectable, however (p = 0.3350).
5. Conclusions In the case of cherry-flavoured beverages or to be more specific in the case of benzaldehyde-containing beverages, industrial best practices should include monitoring for benzene. The International Council of Beverages Associations (2006) guidance
Table 1 Benzene, benzaldehyde, and benzoic acid concentrations of 170 beverages. Cherry-flavoured beverages (n = 72)
Positive >EU benzene water limit (1 lg/l) Average* Standard deviation* Minimum* P90* P95* Maximum* Median* *
Beverages with other flavours (n = 98)
Benzene [lg/L]
Benzaldehyde [mg/L]
Benzoic acid [mg/L]
Benzene [lg/L]
Benzaldehyde [mg/L]
Benzoic acid [mg/L]
54 (75%) 1 (1%) 0.18 0.35 n.d. 0.39 0.53 2.86 0.11
64 (89%) – 8 9 n.d. 20 29 42 5
19 (26%) – 5 21 n.d. 0 20 123 n.d.
19 (19%) 1 (1%) 0.08 0.55 n.d. 0.07 0.09 5.43 n.d.
9 (9%) – 0.1 0.4 n.d. 0.0 0.6 3.8 n.d.
30 (31%) – 18 43 n.d. 114 129 134 n.d.
Distribution calculated with not detectable (n.d.) results below LOD set as zero and values between LOD and LOQ set at the actual numerical value.
C. Loch et al. / Food Chemistry 206 (2016) 74–77
77
Acknowledgments
0.7
The skilful technical assistance of H. Havel and H. Mann is gratefully acknowledged.
0.6 0.5
References 0.4 0.3 0.2 0.1
Linear Fit 95% Prediction Limit R=0.6122 p<0.0001
0.0
0
10
20
30
40
50
Benzaldehyde (mg/L) Fig. 2. Relationship between benzaldehyde and benzene concentrations in a survey of 170 beverages (results below LOD were calculated as zero, two outliers with extreme benzene concentrations were removed).
Table 2 Benzaldehyde and benzoic acid concentrations of benzene-positive and benzene negative samples in a survey of 170 beverages.
Benzene negative samples Benzene positive samples p (Mann–Whitney Test)
Benzaldehyde [mg/L]
Benzoic acid [mg/L]
0.5 ± 1.6 7.5 ± 9.1 <0.0001
11 ± 34 14 ± 38 0.3730
document to mitigate the potential for benzene formation in beverages, for example, has included the check for benzaldehyde as a control point (CP) that beverage developers may wish to consider when formulating a product (similar to the CP for benzoic acid). Formulations containing either benzoic acid or benzaldehyde in combination with ascorbic acid should be avoided. This research suggests that while the awareness about benzoic acid as a CP is high (as documented by the falling levels over the past decade), this may be less the case about benzaldehyde. The industry should implement mitigation measures for benzene in cherry-flavoured beverages in consideration of the ALARA principle. For the beverage and the flavour companies supplying cherry flavourings this demand may present certain problems. The flavouring substance benzaldehyde is an important component of cherry flavourings. However, it may be possible to substitute it with other flavouring substances with cherry/almond notes. This may provide an alternative solution to the benzene problem by eliminating benzaldehyde.
Conflict of interest statement The authors declare that there are no conflicts of interest.
Anon (1984). Bestimmung von Konservierungsstoffen in fettarmen Lebensmitteln. Amtliche Sammlung von Untersuchungsverfahren, L00.00-9, 1–4 (in German). Becalski, A., & Nyman, P. (2009). Benzene. In R. H. Stadler & D. R. Lineback (Eds.), Process-induced food toxicants: occurrence, formation, mitigation, and health risks (pp. 413–444). Hoboken, NJ, USA: John Wiley & Sons Inc. Burdock, G. A. (2009). Fenaroli’s handbook of flavour ingredients (6th ed.) Boca Raton, FL: CRC Press. Cao, X. L., Casey, V., Seaman, S., Tague, B., & Becalski, A. (2007). Determination of benzene in soft drinks and other beverages by isotope dilution headspace gas chromatography/mass spectrometry. Journal of AOAC International, 90, 479–484. European Council (1993). Council Regulation (EEC) No 315/93 laying down Community procedures for contaminants in food. Official Journal of the European Communities, L37, 1–3. European Council (1998). Council Directive 98/83/EC on the quality of water intended for human consumption. Official Journal of the European Communities, L330, 32–54. Gardner, L. K., & Lawrence, G. D. (1993). Benzene production from decarboxylation of benzoic acid in the presence of ascorbic acid and a transition-metal catalyst. Journal of Agricultural and Food Chemistry, 41, 693–695. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans (2012). Benzene. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, 100F, 249–294. International Council of Beverages Associations (2006). ICBA Guidance Document to mitigate the potential for benzene formation in beverages. Brussels, Belgium: UNESDA – Union of European Beverages Associations. Katritzky, A. R., Balasubramanian, M., & Siskin, M. (1990). Aqueous hightemperature chemistry of carbo-and heterocycles. 2. Monosubstituted benzenes: benzyl alcohol, benzaldehyde and benzoic acid. Energy & Fuels, 4, 499–505. Lachenmeier, D. W., Kuballa, T., Reusch, H., Sproll, C., Kersting, M., & Alexy, U. (2010). Benzene in infant carrot juice: further insight into formation mechanism and risk assessment including consumption data from the DONALD study. Food and Chemical Toxicology, 48, 291–297. Lachenmeier, D. W., Reusch, H., Sproll, C., Schoeberl, K., & Kuballa, T. (2008). Occurence of benzene as heat-induced contaminant of carrot juice for babies in a general survey of beverages. Food Additives and Contaminants, 25, 1216–1224. Lachenmeier, D. W., Steinbrenner, N., Löbell-Behrends, S., Reusch, H., & Kuballa, T. (2010). Benzene contamination in heat-treated carrot products including baby foods. The Open Toxicology Journal, 4, 39–42. McNeal, T. P., Nyman, P. J., Diachenko, G. W., & Hollifield, H. C. (1993). Survey of benzene in foods by using headspace concentration techniques and capillary gas-chromatography. Journal of AOAC International, 76, 1213–1219. Nyman, P. J., Diachenko, G. W., Perfetti, G. A., McNeal, T. P., Hiatt, M. H., & Morehouse, K. M. (2008). Survey results of benzene in soft drinks and other beverages by headspace gas chromatography/mass spectrometry. Journal of Agricultural and Food Chemistry, 56, 571–576. Nyman, P. J., Wamer, W. G., Begley, T. H., Diachenko, G. W., & Perfetti, G. A. (2010). Evaluation of accelerated UV and thermal testing for benzene formation in beverages containing benzoate and ascorbic acid. Journal of Food Science, 75, C263–C267. Page, B. D., Conacher, H. B. S., Weber, D., & Lacroix, G. (1992). A survey of benzene in fruits and retail fruit juices, fruit drinks, and soft drinks. Journal of AOAC International, 75, 334–340. Salviano dos Santos, V. P., Medeiros Salgado, A., Guedes Torres, A, & Signori Pereira, K. (2015). Benzene as a chemical hazard in processed foods. International Journal of Food Science Article ID 545640. Steinbrenner, N., Löbell-Behrends, S., Reusch, H., Kuballa, T., & Lachenmeier, D. W. (2010). Benzol in Lebensmitteln – ein Überblick. Journal für Verbraucherschutz und Lebensmittelsicherheit, 5, 443–452 (in German). Stiftung Warentest (2013a). Kritisches Kirscharoma. Test, 7, 30–33 (in German). Stiftung Warentest (2013b). Kunstaroma statt Frucht. Test, 5, 20–26 (in German). Wu, Q. J., Lin, H., Fan, W., Dong, J. J., & Chen, H. L. (2006). Investigation into benzene, trihalomethanes and formaldehyde in Chinese lager beers. Journal of the Institute of Brewing, 112, 291–294.