Abstract
The furan levels in commercial coffee product samples (17 instant coffees, 12 mixed coffee, 8 canned coffee) were 49–2155, 10–201 and 15–209 ng/g, respectively. Since thermal degradation/rearrangement of carbohydrates is the main source of furan, the concentrations of furan and monosaccharides (mannose, rhamnose, glucose, galactose and arabinose) were analysed in 26 green and roasted coffee bean (Coffea arabica) varieties. In coffee beans, furan levels ranged from 4.71 (Bourbon Cerrado, Brazil) to 8.63 mg/kg (San Vicente, Honduras). Galactose was the main monosaccharide in green beans, followed by arabinose, glucose, mannose and rhamnose, on average. Roasting decreased the glucose content by about 81%, and arabinose decreased about 27% in all coffee beans. Glucose decreased the greatest after roasting and is thereby considered the major contributor to the formation of furan.
Keywords: Furan, Coffee, Monosaccharide, Coffee bean variety, Commercial coffee product
Introduction
Recent concerns about the food safety of toxic compounds have increased due to the discovery of numerous, potentially cancer-causing substances. These carcinogenic compounds are found in food ingredients or generated during cooking, such as boiling or frying (Crews and Castle 2007).
Furan is a volatile toxic compound with a boiling point of 31 °C. It is classified as a Group 2B compound (possibly carcinogenic to humans) by the International Agency for Research on Cancer (IARC 1995). International organisations, such as the United States Food and Drugs Administration (FDA) and the European Food Safety Authority (EFSA), have studied the toxicity and risk assessment of furan (Mesías et al. 2013). Furan is formed by numerous pathways during thermal treatment, including the thermal degradation and rearrangement of carbohydrates either alone or in the presence of amino acids (Maillard reaction), the oxidation of polyunsaturated fatty acids and ascorbic acid at high temperature (Nie et al. 2013b), and the thermal degradation of some amino acids (serine and cysteine) to glycolaldehyde and acetaldehyde that can undergo aldol condensation to form furan. Other amino acids (e.g., aspartic acid, α-alanine and threonine) only produce acetaldehyde and so require carbohydrates as a source of glycolaldehyde to form furan. However, carbohydrates are considered the main source of furan formation (Maga and Katz 1979), and ascorbic acid is known to increase the probability of furan formation (Perez Locas and Yaylayan 2004).
Furan occurs in a variety of thermally processed foods, such as canned vegetables, baby foods, fruit preserves, fruit juice, meat, fish and coffee (Food and Administration 2004; Jeong et al. 2019). The consumption of coffee is a major contributor to furan exposure to adults (Han et al. 2017; Seok et al. 2020). Due to temperatures exceeding 200 °C in the roasting process, furan is found in roasted coffee in amounts over 5 μg/g (EFSA 2011).
Coffee is one of the most consumed beverages in the world. A typical coffee beverage is prepared by roasting, grinding and extraction of the coffee beans. During roasting, diverse chemical reactions, such as the Maillard reaction, caramelisation and Strecker degradation could be occurred (Lee et al. 2017). These reactions affect various sensory qualities of coffee, like colour, flavour and aroma (Cha et al. 2019). In addition, a wide range of toxic volatiles, such as furan and 5-methyl furfurals, are formed (Seok et al. 2020). Even for the same type of coffee sample, the levels of furan can differ dramatically due to the coffee beans (Han et al. 2017).
In this study, the levels of furan in the commercial coffee products were monitored to investigate the relationship between furan and monosaccharides in coffee samples. Additionally, the levels of furan and monosaccharides were analysed in 26 varieties of coffee beans (green beans and coffee beans). Through this study various reduction methods for furan in coffee samples could be proposed.
Materials and methods
Chemical reagents and materials
Furan (+ 99%, purity), d4-furan (98% purity), mannose, rhamnose, glucose, galactose and arabinose were purchased from Sigma–Aldrich Corporation (St. Louis, MO, USA). HCl solution was bought from Samchun (Seoul, Korea). HPLC-grade acetone, hexane, methanol, water and chloroform were procured from J.T. Baker (Phillipsburg, NJ, USA).
Sampling and sample preparation
A total of 37 coffee products were purchased from markets or cafés in Korea. An additional 26 varieties of green beans were bought from a commercial market (Triniti Co., Goyang-si, Korea), all of which were Coffea arabica. Green beans (50 g) were roasted in a roaster (CRB-101A, Gene Café, Seoul, Korea) at 240 °C for 15 min just before analysis. Coffee beans and green beans were ground using a grinder (HMF-995, Hanil Electric Co., Seoul, Korea) and analysed as well.
Furan analysis
Furan and d4-furan were stored in a refrigerator and freezer, respectively. The stock solution was prepared by diluting 1 μL aliquot of native furan and d4-furan separately in methanol to give a total volume of 10 mL (100 μg/mL each). Fresh working solutions (100 μL of the respective stock solution diluted with water to give a total volume of 10 mL) were prepared in 20-mL vials (1 μg/mL), sealed with silicon–PTFE septa and aluminium seals and kept on ice.
Sample preparation differed based on the state (liquid or solid) of the material to be analysed. For solid samples, 1 g of sample was placed in a 20-mL headspace vial appropriate for automated solid-phase microextraction (SPME) using a multipurpose sampler (Gerstel GmbH & Co. KG, Mülheim an der Ruhr, Germany), and 9 mL water was added. For liquid samples, 5 mL each of sample and water was placed in a 20-mL headspace vial. The vials were completely closed with silicon–PTFE septa and aluminium seal. An aliquot of 10 μL of internal standard working solution (d4-furan) was spiked into the vials (final concentration of internal standard is 1 ng/mL). Vials were homogenised using a vortex mixer and stored in an icebox filled with ice before analysis.
Furan was extracted from the samples using automated headspace–solid-phase microextraction (HS–SPME). Previous research has already shown that automated HS–SPME provides precise data for furan analysis (Altaki et al. 2007). The HS–SPME procedure was conducted using a carboxen/polydimethylsiloxane (CAR/PDMS) fibre (75 μm thickness; Supelco, Bellefonte, PA, USA). During extraction, the samples were agitated (50 °C, 300 rpm) to encourage the transfer of furan from the matrix to the headspace, and the fibre was exposed to the headspace of the sample vial at 25-mm depth for 20 min. After extraction, the fibre was thermally desorbed in the injection port of a gas chromatograph (GC; 7820A, Agilent Technologies, Santa Clara, CA, USA) at 50 mm depth for 300 s. A fibre bake out was conducted between samples to reduce contamination.
The GC was coupled to a mass spectrometer (MS; 5977E, Agilent Technologies, Santa Clara) and equipped with a HP-PLOT Q capillary column (15 m × 0.32 mm i.d., 20 μm film thickness; J&W Scientific, Folsom, CA, USA). The GC oven temperature was set at 50 °C, held for 2 min, then increased at 25 °C/min to 230 °C and held for 5 min, followed by a post-run at 230 °C for 5 min. The carrier gas was high-purity helium (> 99.999%) with a flow rate of 1.5 mL/min. The GC inlet temperature was set at 250 °C, and the injection mode was split-less. The MS transfer line temperature was set at 250 °C. The MS was operated in positive electron impact ionisation mode (EI+, 70 eV) and selective ion monitoring (SIM) mode. For SIM mode, the current of the following ions was recorded: m/z 68 [M]+ (quantifier) and 39 [M-CHO]+ (qualifier) for furan; m/z 72 [M]+ (quantifier) and 42 [M-CHO]+ (qualifier) for d4-furan.
Analysis of monosaccharides
Hydrolysis of monosaccharides in coffee beans was carried out by placing 0.3 g of sample in a 20-mL vial. After adding 5 mL HCl (1 M), the vial was placed in a water bath at 80 °C for 180 min. The hydrolysed sample was filtered through Whatman No. 2 paper, and 1 mL of the eluate was moved through a pre-conditioned cartridge (Sep Pak C18, Agilent, USA) and placed in a volumetric flask. The flask was filled with HPLC-grade water to 5 mL.
Derivatisation of monosaccharides was undertaken by mixing 100 μL of solution with 100 μL NaOH (0.6 M) in a 2-mL vial. Then, 50 μL of the mixture was mixed with 50 μL of 0.5 M 1-phenyl-3-methyl-5-pyrazolone solution in a 5-mL V-vial using a vortex mixer. The V-vial was placed in a thermos-hygrostat at 70 °C for 100 min. To terminate the reaction, the mixture was cooled to room temperature and neutralised with 50 μL HCl (0.3 M). Next, 1 mL each of water and chloroform was added to the mixture and vortexed. The chloroform layer was removed, and the liquid–liquid extraction process was repeated three times. For HPLC analysis, the aqueous layer was filtered through a 0.45-μm membrane.
The monosaccharides were quantified by a HPLC system (Agilent 1200 Series, Agilent Technologies, Waldbronn, Germany) equipped with a Zorbax Eclipse XDB-C18 column (250 mm × 4.6 mm, 5 μm particle size; Agilent Technologies, Palo Alto, CA, USA) at 30 °C. The two mobile phase solvents (acetonitrile and 0.1 M sodium phosphate buffer, pH 6.7) were used in isocratic mode at 83:17 v/v. The injection volume was 10 μL, and the flow rate was 1 mL/min. Detection was by UV absorbance at 245 nm.
Validation for furan and monosaccharides analysis
The method for determination of furan and monosaccharides (mannose, rhamnose, glucose, galactose, arabinose) was validated by assessing the following parameters: linearity, the limit of detection (LOD), the limit of quantitation (LOQ), recovery and precision. Linearity was calculated from the calibration curve of furan as eight concentration points (0, 1, 2, 5, 10, 20, 50, 100 ng) and each of the monosaccharides as five concentration points (0.5, 1, 2, 10, 20 mmol/kg). The LOD (3.3 × SD/slope of calibration curve) and LOQ (10 × SD/slope of calibration curve) were calculated using water samples. Recovery was performed at three different concentration points. Both the inter-day and intra-day precision were reported as relative standard deviation (RSD, %).
Statistical analysis
All results are the averages of triplicate treatments for each experiment. Differences between samples were determined by variance analysis (ANOVA) and Duncan’s multiple range test using IBM SPSS Statistics 21 (IBM, Chicago, IL, USA). Statistical significance was set at p < 0.05.
Results and discussion
Validation for furan and monosaccharides analysis
The validation results of the method for furan analysis are shown in Table 1. The calibration curve was performed at eight concentration points of furan, and the analysis was tested three times at each point. The curve showed high linearity. The coefficient of determination (R2) was 0.999. The LOD was 0.02 ng/g, and the LOQ was 0.06 ng/g. Recovery ranged from 87.76 to 104.85%. Inter-day precision (RSD, %) and intra-day precision (RSD, %) had ranges of 6.34–8.46% and 0.80–3.50%, respectively.
Table 1.
Validation for furan analysis
| Linearity | LOD (ng/g) | LOQ (ng/g) | |
|---|---|---|---|
| Equation | R2 | ||
| y = 0.088x + 0.3015 | 0.999 | 0.02 | 0.06 |
| Concentration (ng/g) | Recovery (%) | Inter-day (RSD, %) | Intra-day (RSD, %) |
|---|---|---|---|
| 10 | 87.76 | 8.46 | 3.50 |
| 50 | 104.85 | 6.34 | 0.80 |
| 100 | 93.69 | 6.53 | 1.04 |
The validation results of the method for the determination of monosaccharides were shown in Table 2. The R2, LOD and LOQ ranges were 0.9978–0.9998. 0.04–0.10 mmol/kg and 0.11–0.30 mmol/kg, respectively. Recoveries ranged from 86.31 to 116.19%. The inter-day precision ranged from 1.86 to 15.19%, and the intra-day precision ranged from 2.10 to 15.15%.
Table 2.
Validation for monosaccharide analysis
| Monosaccharide | Linearity | LOD (mmol/kg) | LOQ (mmol/kg) | Concentration (mmol/kg) | Accuracy | Precision | ||
|---|---|---|---|---|---|---|---|---|
| Equation | R2 | Recovery (%) | Inter-day (RSD, %) | Intra-day (RSD, %) | ||||
| Mannose | y = 106.01x − 22.234 | 0.9998 | 0.04 | 0.12 | 1 | 113.73 | 6.76 | 2.10 |
| 2 | 97.33 | 1.93 | 5.86 | |||||
| 10 | 100.40 | 7.96 | 2.12 | |||||
| Rhamnose | y = 97.187x − 24.548 | 0.9997 | 0.06 | 0.18 | 1 | 116.19 | 11.02 | 3.28 |
| 2 | 98.27 | 1.86 | 6.35 | |||||
| 10 | 102.16 | 8.52 | 2.65 | |||||
| Glucose | y = 85.369x − 34.281 | 0.9992 | 0.05 | 0.14 | 1 | 114.96 | 15.19 | 3.79 |
| 2 | 94.97 | 7.70 | 10.20 | |||||
| 10 | 101.94 | 9.51 | 3.70 | |||||
| Galactose | y = 73.241x − 33.251 | 0.9984 | 0.10 | 0.30 | 1 | 110.05 | 11.61 | 8.87 |
| 2 | 90.08 | 8.03 | 15.15 | |||||
| 10 | 86.31 | 12.74 | 7.09 | |||||
| Arabinose | y = 115.87x − 81.068 | 0.9978 | 0.04 | 0.11 | 1 | 115.41 | 10.09 | 4.01 |
| 2 | 97.16 | 9.37 | 12.20 | |||||
| 10 | 101.88 | 7.67 | 3.86 | |||||
Analysis of furan and monosaccharides
Analysis of furan in coffee products
A total of 37 coffee products, including instant coffee (17), coffee mix (12) and liquid coffee (8), that is, canned coffee (4) and Americano (4), were analysed for furan. The maximum furan level detected was 2155.47 ng/g in instant coffee, and the lowest was 10.87 ng/g in coffee mix, respectively.
The furan content in instant coffee, coffee mix and liquid coffee (canned coffee, Americano) were shown in Table 3. Furan levels in instant coffee varied significantly from 49.93 ± 0.77 to 2155.47 ± 42.39 ng/g. Sample I_10 had a low level of furan when compared with previous values of 200–700 and 2000–2200 ng/g in instant coffee (Kuballa et al. 2005). Coffee mix samples contained maximal and minimal furan concentrations of 201.91 ± 30.08 and 10.87 ± 1.38 ng/g, respectively, with slight differences between samples. Several of these samples showed higher levels of furan than reported in a coffee mix with different ingredients, which contained between 22.5 and 110 ng/g (Nie et al. 2013a). There is a difference in the levels of furan between instant coffee and coffee mix. This difference may be due to the coffee beans and the process such as roasting. In the case of coffee mix, the amount of coffee in the sample used for analysis would be affected by other ingredients, such as sugar and cream, and their ratio in the product. Therefore, the amount of coffee analysed (weight basis) was decreased, and the levels of furan in the coffee mix may be reduced.
Table 3.
Furan levels in instant coffee, coffee mix and liquid coffee (canned coffee, Americano)
| Type | Sample | Furan (ng/g, ng/mL) |
|---|---|---|
| Instant coffee | I_1 | 139.13 ± 6.72b |
| I_2 | 214.61 ± 8.61c,d | |
| I_3 | 1197.96 ± 66.39j | |
| I_4 | 609.56 ± 25.43h | |
| I_5 | 161.21 ± 2.86b,c | |
| I_6 | 405.78 ± 5.26f | |
| I_7 | 1345.70 ± 17.37k | |
| I_8 | 2155.47 ± 42.39m | |
| I_9 | 517.82 ± 22.60g | |
| I_10 | 49.93 ± 0.77a | |
| I_11 | 831.13 ± 60.63i | |
| I_12 | 870.64 ± 54.69i | |
| I_13 | 206.05 ± 5.38c,d | |
| I_14 | 276.31 ± 4.55e | |
| I_15 | 276.20 ± 4.44e | |
| I_16 | 1715.24 ± 53.84l | |
| I_17 | 259.37 ± 6.39d,e | |
| Coffee mix | M_1 | 22.19 ± 0.63a,b |
| M_2 | 81.11 ± 11.19e | |
| M_3 | 161.51 ± 11.27f | |
| M_4 | 42.80 ± 5.97b,c,d | |
| M_5 | 182.57 ± 26.00f,g | |
| M_6 | 63.47 ± 6.68d,e | |
| M_7 | 39.30 ± 1.29b,c | |
| M_8 | 65.50 ± 5.44d,e | |
| M_9 | 60.32 ± 9.64c,d,e | |
| M_10 | 201.91 ± 30.08g | |
| M_11 | 10.87 ± 1.38a | |
| M_12 | 12.83 ± 1.92a | |
| Canned coffee | C_1 | 15.49 ± 2.49a |
| C_2 | 27.34 ± 2.14b | |
| C_3 | 36.64 ± 0.90c | |
| C_4 | 99.35 ± 6.82d | |
| Americano | A_1 | 77.35 ± 3.79a |
| A_2 | 144.59 ± 23.11b | |
| A_3 | 209.09 ± 8.89c | |
| A_4 | 169.27 ± 16.51b |
1. All values are shown as mean ± standard deviation
2. Different letters show significant mean differences between samples by Duncan’s test (p < 0.05)
3. In case of canned and Americano coffee unit of concentration is ng/mL
Canned coffee contained less furan (15.49 ± 2.49–99.35 ± 6.82 ng/mL) than the Americano coffee (77.35 ± 3.79–209.09 ± 8.89 ng/mL). There was a significant difference between the furan concentrations of each type of product, but it was not as great as the furan level difference by type of instant coffee and coffee mix. According to a previous analysis of canned coffee in Korea, 25.27 and 30.88 ng/mL of furan were detected (Han et al. 2017). These levels of furan were similar to those found in the samples coded C_2 and C_3 in this study. Waizenegger (Waizenegger et al. 2012) showed the furan contents in ‘ready to drink coffee’ were highly diverse (2.4–108.8 mg/kg). All Americano samples were consistent with the results of the previous report (Waizenegger et al., 2012) in which the levels of furan varied between 77.35–209.09 ng/g and 44.5 and 234 mg/kg, respectively. One of the reasons why the levels of furan in Americano differ is that each coffee store uses a different coffee machine when preparing coffee products. In addition, the ratio of coffee powder/beans to water used can lead to different levels of furan (Waizenegger et al. 2012).
There was a significant difference in the furan contents between products of the same type. Processing factors are known to affect furan formation. Some major factors that contribute to the formation of furan during coffee processing are the degree of roasting and grinding of the coffee beans. Higher amounts of furan are found in coffee with a higher degree of roast. Differences in furan content can also be attributed to the type of coffee bean (Guenther et al. 2010).
Analysis of furan in coffee beans
To investigate the effect of coffee beans on the levels of furan in coffee products, the furan contents in 26 different varieties of coffee beans (green and roasted) were analysed, and the data are shown in Table 4. No furan was detected in green beans, but furan was formed through roasting, with levels ranging from 4708.21 ± 136.28 ng/g in Bourbon Cerrado (Brazil) to 8634.75 ± 188.59 ng/g in Vicenta (Honduras), amounting to a difference of about 4000 ng/g.
Table 4.
Furan levels in coffee beans
| Region | Variety | Furan (ng/g) |
|---|---|---|
| Bolivia | Mt Andes | 6700.09 ± 182.32e |
| Bolivia | Segunda | 6727.12 ± 156.45e,f |
| Brazil | Bourbon Cerrado | 4708.21 ± 136.28a |
| Brazil | Nossa Senhora Aparecida | 7427.35 ± 77.22g |
| Burundi | Kay AA | 5813.46 ± 323.10c,d |
| Colombia | Huila | 6120.74 ± 172.91d |
| Colombia | Limu | 6201.16 ± 208.02d |
| Colombia | Los Naranjos | 6948.94 ± 370.75e,f,g |
| Costa Rica | Zarcero Honey | 7229.26 ± 278.18f,g |
| Costa Rica | Don Mayo | 7248.21 ± 351.78g |
| Costa Rica | Don Maya Natural | 7446.66 ± 375.90g |
| Costa Rica | El Alto | 8133.97 ± 198.55h |
| El Salvador | Peaberry | 8230.86 ± 477.64h |
| Ethiopia | Yirgacheffe | 5172.97 ± 84.47a,b |
| Ethiopia | Sidamo Peaberry | 5803.13 ± 353.31c,d |
| Ethiopia | Koke Honey | 6180.53 ± 572.96d |
| Ethiopia | Momora washed specialty | 6185.17 ± 108.15d |
| Ethiopia | Chelba Honey | 7226.39 ± 562.72f,g |
| Guatemala | Santa Lucia | 5547.42 ± 189.90b,c |
| Honduras | Vicenta | 8634.75 ± 188.59h |
| Indonesia | Gold Mandheling | 5753.30 ± 90.57c,d |
| Kenya | Ndurutu | 5743.16 ± 218.91c,d |
| Kenya | Gaki | 8451.75 ± 155.93h |
| Nicaragua | Jose | 5080.65 ± 96.83a,b |
| Rwanda | Ahnzu | 6025.19 ± 206.62c,d |
| Tanzania | Momo | 6052.68 ± 320.28c,d |
1. All values are shown as mean ± standard deviation
2. Different letters show significant mean differences between samples by Duncan’s test (p < 0.05)
A previous study detected between 911 and 5852 ng/g of furan in coffee beans (Pavesi Arisseto et al. 2011; Jeong et al. 2019; Chain et al. 2017), which is lower than the levels of furan in this study. The relatively higher amounts found in this study are because the analysed coffee beans were roasted at a higher temperature for a longer time.
Analysis of monosaccharides in green and roasted coffee beans
Analytical data of the levels of five monosaccharides quantified in the green and roasted coffee beans (Table 5) revealed 64.98–134.23 and 134.38–381.28 mmol/kg of mannose, 29.20–72.48 and 28.93–43.96 mmol/kg of rhamnose, 198.53–324.60 and 44.31–51.99 mmol/kg of glucose, 350.40–455.72 and 277.19–492.14 mmol/kg of galactose, and 222.30–323.40 and 173.07–222.54 mmol/kg of arabinose, respectively. These differences between the monosaccharides in green beans and coffee beans were significant. On average, galactose was the main monosaccharide, and rhamnose was the minor monosaccharide, while the amounts of the three remaining monosaccharides depended on whether the sample was green beans (arabinose > glucose > mannose) or roasted beans (mannose > arabinose > glucose). Monosaccharides are known as one of main precursors for the formation of furan. The correlation between the levels of furan and monosaccharides is very low. Their r2 value was significantly lower than 0.1, indicating almost no correlation.
Table 5.
Levels of monosaccharides (mmol/kg) in green beans and roasted beans
| Region | Variety | Mannose | Rhamnose | Glucose | Galactose | Arabinose | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Green beans | Roasted beans | Green beans | Roasted beans | Green beans | Roasted beans | Green beans | Roasted beans | Green beans | Roasted beans | ||
| Bolivia | Mt Andes | 88.50 ± 17.92 | 381.28 ± 16.36 | 30.57 ± 0.90 | 32.27 ± 0.52 | 202.05 ± 21.00 | 47.56 ± 1.24 | 364.36 ± 18.05 | 492.14 ± 18.58 | 228.06 ± 19.71 | 183.97 ± 3.88 |
| Bolivia | Segunda | 88.91 ± 17.20 | 370.68 ± 20.10 | 30.16 ± 0.24 | 31.17 ± 0.74 | 199.38 ± 17.06 | 49.10 ± 0.70 | 403.18 ± 16.67 | 467.01 ± 28.79 | 241.82 ± 16.59 | 189.69 ± 10.04 |
| Brazil | Nossa Senhora Aparecida | 122.50 ± 11.84 | 191.81 ± 13.47 | 30.16 ± 0.24 | 31.76 ± 3.16 | 279.32 ± 23.69 | 46.37 ± 0.32 | 401.87 ± 35.76 | 301.57 ± 7.80 | 265.49 ± 13.52 | 198.30 ± 4.98 |
| Brazil | Bourbon Cerrado | 78.53 ± 6.75 | 334.27 ± 14.82 | 30.11 ± 1.90 | 32.57 ± 0.74 | 198.53 ± 38.62 | 49.68 ± 0.59 | 364.98 ± 25.00 | 466.23 ± 22.07 | 226.43 ± 27.68 | 198.45 ± 6.95 |
| Burundi | Kay AA | 85.78 ± 3.12 | 224.02 ± 11.04 | 47.53 ± 3.62 | 33.76 ± 3.38 | 324.60 ± 29.48 | 49.28 ± 0.97 | 455.72 ± 14.57 | 344.40 ± 3.34 | 323.40 ± 14.31 | 195.39 ± 2.38 |
| Colombia | Los Naranjos | 129.54 ± 13.38 | 189.15 ± 10.94 | 67.33 ± 2.66 | 34.91 ± 1.64 | 284.99 ± 34.57 | 46.45 ± 2.22 | 395.04 ± 42.79 | 280.64 ± 19.12 | 276.52 ± 22.83 | 179.21 ± 8.49 |
| Colombia | Huila | 118.21 ± 13.74 | 214.19 ± 16.17 | 49.55 ± 4.64 | 34.01 ± 5.27 | 228.12 ± 24.45 | 50.47 ± 1.74 | 373.36 ± 48.05 | 359.42 ± 31.12 | 259.85 ± 20.19 | 202.04 ± 12.06 |
| Colombia | Limu | 87.77 ± 19.97 | 334.04 ± 9.62 | 29.77 ± 0.97 | 31.79 ± 1.70 | 244.98 ± 29.29 | 48.46 ± 0.14 | 394.15 ± 16.50 | 454.32 ± 2.44 | 235.03 ± 17.83 | 175.75 ± 0.58 |
| Costa Rica | El Alto | 120.27 ± 12.08 | 201.13 ± 11.49 | 58.65 ± 4.63 | 34.22 ± 4.05 | 250.78 ± 20.49 | 45.62 ± 0.99 | 412.66 ± 24.74 | 312.23 ± 12.35 | 261.71 ± 9.25 | 186.44 ± 7.47 |
| Costa Rica | Don Maya Natural | 107.19 ± 10.42 | 190.11 ± 4.97 | 56.16 ± 4.04 | 31.18 ± 2.64 | 235.78 ± 21.08 | 46.22 ± 0.21 | 369.60 ± 35.17 | 333.37 ± 3.53 | 261.51 ± 18.04 | 199.67 ± 2.40 |
| Costa Rica | Zarcero Honey | 102.71 ± 8.43 | 210.78 ± 7.87 | 39.24 ± 0.55 | 32.06 ± 2.49 | 246.45 ± 3.34 | 45.95 ± 0.41 | 375.89 ± 18.38 | 340.08 ± 9.89 | 280.89 ± 5.57 | 194.13 ± 4.91 |
| Costa Rica | Don Mayo | 64.98 ± 6.14 | 134.38 ± 16.92 | 29.45 ± 1.51 | 32.98 ± 2.85 | 201.11 ± 26.97 | 44.31 ± 1.33 | 422.85 ± 35.42 | 284.42 ± 12.83 | 232.79 ± 5.26 | 181.10 ± 6.14 |
| El Salvador | Peaberry | 74.47 ± 4.92 | 368.70 ± 25.08 | 29.20 ± 0.19 | 31.70 ± 2.66 | 216.15 ± 35.54 | 48.53 ± 0.72 | 364.72 ± 37.33 | 475.00 ± 18.77 | 222.30 ± 18.49 | 173.07 ± 4.37 |
| Ethiopia | Momora washed specialty | 111.06 ± 8.32 | 197.68 ± 15.46 | 45.28 ± 1.72 | 34.57 ± 4.38 | 252.01 ± 14.48 | 44.44 ± 0.47 | 411.12 ± 17.04 | 323.65 ± 17.91 | 303.94 ± 15.16 | 195.89 ± 6.40 |
| Ethiopia | Sidamo Peaberry | 114.89 ± 10.55 | 175.22 ± 1.19 | 67.91 ± 4.72 | 33.23 ± 3.26 | 243.56 ± 33.18 | 46.94 ± 0.41 | 406.24 ± 48.36 | 334.06 ± 2.79 | 289.54 ± 21.72 | 222.54 ± 5.01 |
| Ethiopia | Koke Honey | 134.23 ± 4.37 | 194.77 ± 17.51 | 72.48 ± 3.93 | 43.96 ± 1.31 | 268.95 ± 9.33 | 48.44 ± 0.63 | 443.24 ± 11.99 | 344.85 ± 8.58 | 295.14 ± 6.87 | 209.80 ± 2.18 |
| Ethiopia | Yirgacheffe | 123.90 ± 6.99 | 210.11 ± 3.08 | 61.58 ± 9.66 | 32.90 ± 2.54 | 236.17 ± 29.58 | 49.09 ± 0.55 | 412.03 ± 15.68 | 363.06 ± 19.42 | 280.46 ± 14.33 | 207.86 ± 9.56 |
| Guatemala | Santa Lucia | 118.72 ± 14.95 | 194.78 ± 15.15 | 59.76 ± 5.67 | 33.77 ± 3.65 | 264.81 ± 31.37 | 45.64 ± 0.67 | 416.93 ± 34.54 | 305.45 ± 14.35 | 269.13 ± 19.19 | 190.76 ± 4.63 |
| Honduras | Vicenta | 98.87 ± 7.75 | 211.56 ± 5.34 | 42.69 ± 1.30 | 34.10 ± 2.44 | 272.11 ± 16.40 | 46.68 ± 0.93 | 376.42 ± 13.27 | 337.79 ± 14.73 | 270.91 ± 3.26 | 201.82 ± 7.04 |
| Indonesia | Gold Mandheling | 78.48 ± 7.17 | 340.14 ± 21.97 | 29.66 ± 0.79 | 32.88 ± 2.34 | 252.96 ± 10.18 | 51.99 ± 0.89 | 370.30 ± 51.11 | 487.45 ± 14.11 | 236.56 ± 44.05 | 200.50 ± 4.46 |
| Kenya | Gaki | 111.47 ± 8.72 | 197.75 ± 12.33 | 39.63 ± 4.59 | 28.93 ± 0.75 | 264.35 ± 7.87 | 46.26 ± 0.82 | 350.40 ± 25.65 | 277.19 ± 17.86 | 263.63 ± 11.75 | 180.33 ± 7.05 |
| Kenya | Ndurutu | 113.32 ± 19.85 | 214.46 ± 5.81 | 48.22 ± 4.85 | 39.38 ± 5.81 | 298.04 ± 33.38 | 47.47 ± 1.02 | 390.90 ± 39.95 | 317.99 ± 17.93 | 284.13 ± 25.37 | 187.78 ± 8.04 |
| Nicaragua | Jose | 99.56 ± 4.73 | 166.68 ± 2.27 | 42.58 ± 1.29 | 30.63 ± 0.57 | 211.69 ± 20.76 | 46.56 ± 0.45 | 392.39 ± 17.53 | 314.80 ± 1.11 | 279.91 ± 14.89 | 205.45 ± 2.13 |
| Rwanda | Ahnzu | 122.83 ± 17.69 | 215.50 ± 35.98 | 54.45 ± 2.93 | 37.98 ± 5.87 | 271.87 ± 10.08 | 49.11 ± 2.09 | 395.14 ± 37.72 | 340.24 ± 32.87 | 275.12 ± 17.33 | 208.87 ± 7.97 |
| Tanzania | Momo | 118.50 ± 23.31 | 193.54 ± 7.41 | 44.68 ± 2.37 | 30.58 ± 0.33 | 282.82 ± 43.94 | 46.04 ± 1.09 | 402.09 ± 51.42 | 302.70 ± 12.09 | 277.77 ± 38.29 | 198.90 ± 5.90 |
1. All values are shown as mean ± standard deviation
2. The unit of concentration is mmol/kg
3. Different letters show significant mean differences between samples by Duncan’s test (p < 0.05)
After roasting, the levels of glucose and arabinose decreased in all samples. Glucose decreased by 81.04% on average, the largest decline among the monosaccharides, and arabinose decreased by an average of 26.80%. Roasting increased the mannose content in all samples by an average of 122.86%. Rhamnose and galactose contents increased in Mt Andes (Bolivia), Segunda (Bolivia), Bourbon Cerrado (Brazil), Limu (Colombia), Peaberry (El Salvador) and Gold Mandheling (Indonesia) but decreased in the remaining 20 samples. Except for the levels of rhamnose in Limu (Colombia), Don Mayo (Costa Rica) and Peaberry (El Salvador), the levels of monosaccharides in all samples differed significantly after roasting.
After roasting, when furan was formed, levels of glucose decreased most significantly. Glucose is considered to be the major contributor to the formation of furan among monosaccharides. Relevant research on the effect of monosaccharides has shown that notable amounts of furan are generated during the thermal decomposition of sugars alone, known as caramelisation (Owczarek-Fendor et al. 2012). Furthermore, the presence of the amino acids alanine, threonine and serine, increases the amount of furan, especially when combined with glucose (Limacher et al. 2008).
Conclusion
In this study, the furan contents in commercial coffee products, as well as 26 varieties of coffee beans roasted under the same conditions, were analysed. Monosaccharides, which are known as precursors of furan, were quantified in green and roasted coffee beans. Furan was detected in commercial coffee products at levels of 10.87–2155.47 ng/g, and in coffee beans at 4708.21–8634.75 ng/g, respectively. Green beans showed monosaccharide levels decreased in the order of galactose, arabinose, glucose, mannose and rhamnose, on average. In all coffee beans, roasting decreased the glucose content by about 81.04%, and arabinose decreased by about 26.80%. In most samples, the levels of monosaccharides differed significantly after roasting. Three was no correlation between the levels of furan and those of the analysed monosaccharides. However, glucose is considered to be the major contributor to the formation of furan because its level decreased significantly after roasting.
The results of this study can be used as reference data for the selection of coffee beans based on the levels of furan in 26 varieties of coffee beans. It is meaningful to analyse the levels of monosaccharide in various varieties of green beans and coffee beans. Future research on other components known as the precursors of furan in green beans and coffee beans is needed.
Acknowledgements
This research was supported by the Korea Institute of Planning and Evaluation for Technology in Food, Agriculture, Forestry and Fisheries (IPET 118012031SB010); the Agriculture, Food and Rural Affairs Research Centre Support Program, through the Ministry of Agriculture, Food and Rural Affairs; and the Basic Science Research Program through the National Research Foundation of Korea (NRF 018R1A2B6002634).
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- Altaki M, Santos F, Galceran M. Analysis of furan in foods by headspace solid-phase microextraction–gas chromatography–ion trap mass spectrometry. J Chromatogr A. 2007;1146(1):103–109. doi: 10.1016/j.chroma.2007.01.104. [DOI] [PubMed] [Google Scholar]
- Cha J, Debnath T, Lee K-G. Analysis of α-dicarbonyl compounds and volatiles formed in Maillard reaction model systems. Sci Rep. 2019;9(1):1–6. doi: 10.1038/s41598-018-37186-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chain EPOCITF, Knutsen HK, Alexander J, Barregård L, Bignami M, Brüschweiler B, Ceccatelli S, Cottrill B, Dinovi M, Edler L. Risks for public health related to the presence of furan and methylfurans in food. EFSA J. 2017;15(10):e05005. doi: 10.2903/j.efsa.2017.5005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Crews C, Castle L. A review of the occurrence, formation and analysis of furan in heat-processed foods. Trends Food Sci Technol. 2007;18(7):365–372. doi: 10.1016/j.tifs.2007.03.006. [DOI] [Google Scholar]
- EFSA Update on furan levels in food from monitoring years 2004–2010 and exposure assessment. EFSA J. 2011;9(9):2347. [Google Scholar]
- Food U, Administration D (2004) Exploratory data on furan in food. FDA, Washington, DC (May 7, 2004, updated June 7, 2004, June 15, 2005 and October 27, 2006. http://www.cfscanfda.gov/about~dms/furandat.htm1)
- Guenther H, Hoenicke K, Biesterveld S, Gerhard-Rieben E, Lantz I. Furan in coffee: pilot studies on formation during roasting and losses during production steps and consumer handling. Food Addit Contam. 2010;27(3):283–290. doi: 10.1080/19440040903317505. [DOI] [PubMed] [Google Scholar]
- Han J, Kim MK, Lee KG. Furan levels and sensory profiles of commercial coffee products under various handling conditions. J Food Sci. 2017;82(11):2759–2766. doi: 10.1111/1750-3841.13933. [DOI] [PubMed] [Google Scholar]
- IARC Furan. IARC Monogr Eval Carcinog Risks Hum. 1995;63:393–407. [PMC free article] [PubMed] [Google Scholar]
- Jeong SY, Jang HW, Debnath T, Lee KG. Validation of analytical method for furan determination in eight food matrices and its levels in various foods. J Sep Sci. 2019;42(5):1012–1018. doi: 10.1002/jssc.201800736. [DOI] [PubMed] [Google Scholar]
- Kuballa T, Stier S, Strichow N. Furan concentrations in coffee and coffee beverages. Dtsch Lebensm-Rundsch. 2005;101(6):229–235. [Google Scholar]
- Lee SJ, Kim MK, Lee K-G. Effect of reversed coffee grinding and roasting process on physicochemical properties including volatile compound profiles. Innov Food Sci Emerg Technol. 2017;44:97–102. doi: 10.1016/j.ifset.2017.07.009. [DOI] [Google Scholar]
- Limacher A, Kerler J, Davidek T, Schmalzried F, Blank I. Formation of furan and methylfuran by Maillard-type reactions in model systems and food. J Agric Food Chem. 2008;56(10):3639–3647. doi: 10.1021/jf800268t. [DOI] [PubMed] [Google Scholar]
- Maga JA, Katz I. Furans in foods. Crit Rev Food Sci Nutr. 1979;11(4):355–400. doi: 10.1080/10408397909527268. [DOI] [PubMed] [Google Scholar]
- Mesías M, Guerra-Hernández E, García-Villanova B. Furan content in Spanish baby foods and its relation with potential precursors. CyTA J Food. 2013;11(1):1–6. doi: 10.1080/19476337.2012.669797. [DOI] [Google Scholar]
- Nie S-P, Huang J-G, Zhang Y-N, Hu J-L, Wang S, Shen M-Y, Li C, Marcone MF, Xie M-Y. Analysis of furan in heat-processed foods in China by automated headspace gas chromatography–mass spectrometry (HS–GC–MS) Food Control. 2013;30(1):62–68. doi: 10.1016/j.foodcont.2012.07.020. [DOI] [Google Scholar]
- Nie S, Huang J, Hu J, Zhang Y, Wang S, Li C, Marcone M, Xie M. Effect of pH, temperature and heating time on the formation of furan in sugar–glycine model systems. Food Sci Hum Wellness. 2013;2(2):87–92. doi: 10.1016/j.fshw.2013.05.001. [DOI] [Google Scholar]
- Owczarek-Fendor A, De Meulenaer B, Scholl G, Adams A, Van Lancker F, Eppe G, De Pauw E, Scippo M-L, De Kimpe N. Furan formation in starch-based model systems containing carbohydrates in combination with proteins, ascorbic acid and lipids. Food Chem. 2012;133(3):816–821. doi: 10.1016/j.foodchem.2012.01.098. [DOI] [Google Scholar]
- Pavesi Arisseto A, Vicente E, Soares Ueno M, Verdiani Tfouni SAI, De Figueiredo Toledo MCl. Furan levels in coffee as influenced by species, roast degree, and brewing procedures. J Agric Food Chem. 2011;59(7):3118–3124. doi: 10.1021/jf104868g. [DOI] [PubMed] [Google Scholar]
- Perez Locas C, Yaylayan VA. Origin and mechanistic pathways of formation of the parent furan A food toxicant. J Agric Food Chem. 2004;52(22):6830–6836. doi: 10.1021/jf0490403. [DOI] [PubMed] [Google Scholar]
- Seok Y-J, Lee K-G, Biotechnology Analysis of furan in semi-solid and paste type foods. J Food Sci. 2020;29(2):293–301. doi: 10.1007/s10068-019-00654-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Waizenegger J, Winkler G, Kuballa T, Ruge W, Kersting M, Alexy U, Lachenmeier D. Analysis and risk assessment of furan in coffee products targeted to adolescents. Food Addit Contam Part A. 2012;29(1):19–28. doi: 10.1080/19440049.2011.617012. [DOI] [PubMed] [Google Scholar]