Incidence,level, and behavior of aflatoxins during coffee bean roasting and decaffeination

Screening for aflatoxins (Afs), isolation and identification of Aspergillus flavus, and the effect of decaffeination and roasting on the level of contamination in coffee beans are studied. The percent frequency of A. flavus ranged between 4 and 80% in green coffee beans (GCB), whereas in ground roasted coffee beans (GRCB), it ranged between 1 and 71%. Aflatoxins were detected in 76.5 and 54.6% of the infected samples with averages of 4.28 and 2.85 microg/kg of GCB and GRCB, respectively. Roasting was demonstrated to lower the concentration of Afs in GCB. The Afs levels were reduced by approximately 42.2-55.9% depending on the type and temperature of roasting. The highest yields of Afs were detected in the decaffeinated green coffee beans (24.29 microg/kg) and roasted coffee beans (16.00 microg/kg). The growth of A. flavus in liquid medium containing 1 or 2% caffeine was reduced by 50%, and the level of aflatoxin in the medium was undetectable.     

Release kinetics of volatile organic compounds from roasted and ground coffee: online measurements by PTR-MS and mathematical modeling

The present work shows the possibilities and limitations in modeling release kinetics of volatile organic compounds (VOCs) from roasted and ground coffee by applying physical and empirical models such as the diffusion and Weibull models. The release kinetics of VOCs were measured online by proton transfer reaction-mass spectrometry (PTR-MS). Compounds were identified by GC-MS, and the contribution of the individual compounds to different mass fragments was elucidated by GC/PTR-MS. Coffee samples roasted to different roasting degrees and ground to different particle sizes were studied under dry and wet stripping conditions. To investigate the accuracy of modeling the VOC release kinetics recorded using PTR-MS, online kinetics were compared with kinetics reconstituted from purge and trap samplings. Results showed that uncertainties in ion intensities due to the presence of isobaric species may prevent the development of a robust mathematical model. Of the 20 identified compounds, 5 were affected to a lower extent as their contribution to specific m/z intensity varied by <15% over the stripping time. The kinetics of these compounds were fitted using physical and statistical models, respectively, the diffusion and Weibull models, which helped to identify the underlying release mechanisms. For dry stripping, the diffusion model allowed a good representation of the release kinetics, whereas for wet stripping conditions, release patterns were very complex and almost specific for each compound analyzed. In the case of prewetted coffee, varying particle size (approximately 400-1200 microm) had no significant effect on the VOC release rate, whereas for dry coffee, the release was faster for smaller particles. The absence of particle size effect in wet coffee was attributed to the increase of opened porosity and compound diffusivity by solubilization and matrix relaxation. To conclude, the accurate modeling of VOC release kinetics from coffee allowed small variations in compound release to be discriminated. Furthermore, it evidenced the different aroma compositions that may be obtained depending on the time when VOCs are recovered.     

Modeling the secondary shelf life of ground roasted coffee

This work was addressed to study the secondary shelf life of ground roasted coffee. To this purpose, fresh dark-roasted ground coffee samples were equilibrated at increasing water activity (aw) values up to 0.44 and stored at 30 degrees C for up to 1 month. To simulate home storage conditions, the headspace atmosphere was periodically perturbed by opening for a short time and then closing the packaging. The changes of some chemical and physicochemical indexes of coffee staling were studied, and sensory analysis was carried out to determine the end point of coffee acceptability. The results showed that the volatile compounds in the headspace are representative indexes of the quality depletion of roasted ground coffee during home usage. The sensory and instrumental results were used to develop a mathematical model allowing to simply and quickly calculate the secondary shelf life of coffee on the basis of its aw value at a given temperature.     

Alkylpyridiniums. 2. Isolation and quantification in roasted and ground coffees

Recent model studies on trigonelline decomposition have identified nonvolatile alkylpyridiniums as major reaction products under certain physicochemical conditions. The quaternary base 1-methylpyridinium was isolated from roasted and ground coffee and purified by ion exchange and thin-layer chromatography. The compound was characterized by nuclear magnetic resonance spectroscopy ((1)H, (13)C) and mass spectrometry techniques. A liquid chromatography-electrospray ionization tandem mass spectrometry method was developed to quantify the alkaloid in coffee by isotope dilution mass spectrometry. The formation of alkylpyridiniums is positively correlated to the roasting degree in arabica coffee, and highest levels of 1-methylpyridinium, reaching up to 0.25% on a per weight basis, were found in dark roasted coffee beans. Analyses of coffee extracts also showed the presence of dimethylpyridinium, at concentrations ranging from 5 to 25 mg/kg. This is the first report on the isolation and quantification of alkylpyridiniums in coffee. These compounds, described here in detail for the first time, may have an impact on the flavor/aroma profile of coffee directly (e.g., bitterness), or indirectly as precursors, and potentially open new avenues in the flavor/aroma modulation of coffee.     

Analysis of volatile compounds released during the grinding of roasted coffee beans using solid-phase microextraction

A dynamic solid-phase microextraction (SPME) method to sample fresh headspace volatile compounds released during the grinding of roasted coffee beans was described and the analytical results using gas chromatography/mass spectrometry (GC/MS) and GC/olfactometry (GC/O) were compared to those of the conventional static SPME sampling methods using ground coffee. Volatile compounds released during the grinding of roasted coffee beans (150 g) were obtained by exposing the SPME fiber (poly(dimethylsiloxane)/divinylbenzene, PDMS/ DVB) for 8 min to nitrogen gas (600 mL/min) discharged from a glass vessel in which the electronic coffee grinder was enclosed. Identification and characterization of volatile compounds thus obtained were achieved by GC/MS and GC/O. Peak areas of 47 typical coffee volatile compounds, separated on total ion chromatogram (TIC), obtained by the dynamic SPME method, showed coefficients of variation less than 5% (n = 3) and the gas chromatographic profile of volatile compounds thus obtained was similar to that of the solvent extract of ground coffee, except for highly volatile compounds such as 4-hydroxy-2,5-dimethyl-3(2H)-furanone and 4-ethenyl-2-methoxyphenol. Also, SPME dilution analysis of volatile compounds released during the grinding of roasted coffee beans showed linear plots of peak area versus exposed fiber length (R (2) > 0.89). Compared with those of the headspace volatile compounds of ground coffee using GC/MS and GC/O, the volatile compounds generated during the grinding of roasted coffee beans were rich in nutty- and smoke-roast aromas.     

Role of water state and mobility on the antiplasticization of green and roasted coffee beans

The effect of water on "antiplasticization" and plasticization of green and roasted coffee was studied by textural analysis, sorption isotherms, differential scanning calorimetry (DSC), and nuclear magnetic resonance (NMR). From BET monolayer value to a(w) = 0.61 and 0.75 for green and roasted coffee, respectively, the solid matrix hydration occurred and water induced hardening. Very short NMR T(2) values and the concomitant absence of any DSC endothermic peak assignable to water freezing were observed at these a(w) values. When solid matrix hydration was completed, water started to act as a plasticizing agent, the compressive modulus started to decrease, and NMR revealed the appearance of a new proton pool with increased mobility. According to DSC, only when the plasticizing effect became important did water present enough mobility to freeze. Above this moisture value (a(w) = 0.78 and 0.86 for green and roasted coffee, respectively), water determined a decrease of bean hardness and a further decrease of the elastic modulus.     

Effect of roasting on the formation of chlorogenic acid lactones in coffee

Of all plant constituents, coffee has one of the highest concentrations of chlorogenic acids. When roasting coffee, some of these are transformed into chlorogenic acid lactones (CGL). We have studied the formation of CGL during the roasting of coffee beans in Coffea arabica cv. Bourbon; C. arabicacv. Longberry; and C. canephora cv. Robusta. Individual CGL levels were determined by comparison of HPLC peaks with those of synthetic CGL standards. Seven CGL were identified: 3-caffeoylquinic-1,5-lactone (3-CQL), 4- caffeoylquinic-1,5-lactone (4-CQL), 3-coumaroylquinic-1,5-lactone (3-pCoQL), 4-coumaroylquinic-1,5-lactone (4-pCoQL), 3-feruloylquinic-1,5-lactone (3-FQL), 4-feruloylquinic-1,5-lactone (4-FQL), and 3,4-dicaffeoylquinic-1,5-lactone (3,4-diCQL). 3-CQL was the most abundant lactone in C. arabica and C. canephora, reaching peak values of 230 +/- 9 and 254 +/- 4 mg/100 g (dry weight), respectively, at light medium roast ( approximately 14% weight loss). 4-CQL was the second most abundant lactone (116 +/- 3 and 139 +/- 2 mg/100 g, respectively. The maximum amount of CGL represents approximately 30% of the available precursors. The relative levels of 3-CQL and 4-CQL in roasted coffee were reverse to those of their precursors in green coffee. This suggests that roasting causes isomerization of chlorogenic acids prior to the formation of lactones and that the levels of lactones in roasted coffee do not reflect the levels of precursors in green coffee.     

Furan levels in coffee as influenced by species, roast degree, and brewing procedures

Brazilian green coffee beans of Coffea arabica and Coffea canephora species were roasted to light, medium, and dark roast degrees and analyzed in relation to furan content by using an in-house validated method based on gas chromatography coupled to mass spectrometry preceded by headspace solid-phase microextraction. Furan was not detected in green coffees, whereas levels between 911 and 5852 μg/kg were found in the roasted samples. Higher concentrations were found in Coffea canephora species and darker ground coffees. Some of the potential furan precursors were observed in significant amounts in green coffee, especially sucrose and linoleic acid, but their concentrations could not be correlated to furan formation. Additionally, coffee brews were prepared from roasted ground coffees by using two different procedures, and furan levels in the beverages varied from <10 to 288 μg/kg. The factor that most influenced the furan content in coffee brew was the brewing procedure.     

Roasting and aroma formation: effect of initial moisture content and steam treatment

Initial moisture of green coffee may vary as a function of green coffee processing and storage conditions. The impact of initial moisture and steam treatment on roasting behavior and aroma formation was investigated. Steam treated coffees as well as coffees with initial moisture content of 5.10, 10.04, and 14.70 g water per 100 g wb were roasted. Light and dark roasting trials were carried out using a fluidizing-bed roaster with a batch size of 100 g of green beans. Differences in roast coffee attributes, that is, color, density, and organic roast loss, and odorant concentrations were more marked in light roasted than in dark roasted coffees. The results of roasting steam treated coffee suggest that this step affects roasting behavior primarily by extracting some aroma precursor compounds.     

Antioxidant properties of roasted coffee residues

The antioxidant activity of roasted coffee residues was evaluated. Extraction with four solvents (water, methanol, ethanol, and n-hexane) showed that water extracts of roasted coffee residues (WERCR) produced higher yields and gave better protection for lipid peroxidation. WERCR showed a remarkable protective effect on oxidative damage of protein. In addition, WERCR showed scavenging of free radicals as well as the reducing ability and to bind ferrous ions, indicating that WERCR acts as both primary and secondary antioxidants. The HPLC analyses showed that phenolic acids (chlorogenic acid and caffeic acid) and nonphenolic compounds [caffeine, trigonelline, nicotinic acid, and 5-(hydroxymethyl)furfuraldehyde] remained in roasted coffee residues. These compounds showed a protective effect on a liposome model system. The concentrations of flavonoids and polyphenolic compounds in roasted coffee residues were 8,400 and 20,400 ppm, respectively. In addition, the Maillard reaction products (MRPs) remaining in roasted coffee residues were believed to show antioxidant activity. These data indicate that roasted coffee residues have excellent potential for use as a natural antioxidant source because the antioxidant compounds remained in roasted coffee residues.     

Evaluation of methods used to determine ochratoxin A in coffee beans

A comparative study was conducted to evaluate four previously reported methods that proved to have a recovery greater than 80% for the determination of different levels of ochratoxin A (OTA) in green and roasted coffee beans and to select an accurate, sensitive, and less-expensive technique between the existing methods. The results indicated that the Association of Official Analytical Chemists (AOAC) official method for the extraction of OTA in green coffee and determination by high-performance liquid chromatography (HPLC) is recommended as an efficient method for the routine analyses of OTA in green and ground roasted coffee beans. This method proved to be an accurate, sensitive, and less-expensive method that employs routine materials and available equipment. Although the immunoaffinity column/HPLC procedure tested showed a significantly higher percentage than the AOAC recommended method, it is recommended for use in processed coffee beans where low concentrations of OTA may be expected to be detected.     

Identification of proline-based diketopiperazines in roasted coffee

Five proline-based diketopiperazines were identified in water extracts of roasted coffee proteins and roasted coffee itself. These are cyclo(pro-ile), cyclo(pro-leu), cyclo(pro-phe), cyclo(pro-pro), and cyclo(pro-val). The isolation included gel chromatography and solvent (CHCl(3)) extraction; in the case of roasted coffee brews, polyamide column chromatography was also used. The identification was achieved by LC-ESI-MS and -MS/MS by comparison of the retention time and the fragmentation pattern with reference compounds. As a second method GC-EI-MS was used. By both methods the presence of diketopiperazines in roasted coffee was unambiguously verified.     

Interactions between volatile and nonvolatile coffee components. 1. Screening of nonvolatile components

This study is the first of two publications that investigate the phenomena of coffee nonvolatiles interacting with coffee volatile compounds. The purpose was to identify which coffee nonvolatile(s) are responsible for the interactions observed between nonvolatile coffee brew constituents and thiols, sulfides, pyrroles, and diketones. The overall interaction of these compounds with coffee brews prepared with green coffee beans roasted at three different roasting levels (light, medium, and dark), purified nonvolatiles, and medium roasted coffee brew fractions (1% solids after 1 or 24 h) was measured using a headspace solid-phase microextraction technique. The dark roasted coffee brew was slightly more reactive toward the selected compounds than the light roasted coffee brew. Selected pure coffee constituents, such as caffeine, trigonelline, arabinogalactans, chlorogenic acid, and caffeic acid, showed few interactions with the coffee volatiles. Upon fractionation of medium roasted coffee brew by solid-phase extraction, dialysis, size exclusion chromatography, or anion exchange chromatography, characterization of each fraction, evaluation of the interactions with the aromas, and correlation between the chemical composition of the fractions and the magnitude of the interactions, the following general conclusions were drawn. (1) Low molecular weight and positively charged melanoidins present significant interactions. (2) Strong correlations were shown between the melanoidin and protein/peptide content, on one hand, and the extent of interactions, on the other hand (R = 0.83-0.98, depending on the volatile compound). (3) Chlorogenic acids and carbohydrates play a secondary role, because only weak correlations with the interactions were found in complex matrixes.     

Rapid analysis of cholesterol-elevating compounds in coffee brews by off-line high-performance liquid chromatography/high-resolution gas chromatography

A new method is proposed to analyze the cholesterol-elevating cafestol and kahweol which allows their rapid and reliable determination in different coffee brews. The method involves the preseparation of the sample by high-performance liquid chromatography, the collection of the selected fraction, and its subsequent analysis by high-resolution gas chromatography using a programmed temperature vaporizer operated in the split mode as sampling system. Under the experimental conditions investigated, recoveries as high as 87% (cafestol) and 94% (kahweol) were achieved while detection limits equal to 0.06 and 0.04 ppm for cafestol and kahweol, respectively, were obtained. Examples are given comparing levels of cafestol and kahweol resulting from the same ground roasted coffee by different brewing methods, which show the lowest values for brews prepared from coffee bags.     

Chemical characterization of the high-molecular-weight material extracted with hot water from green and roasted robusta coffees as affected by the degree of roast

The hot-water-soluble polymeric material from green and roasted Uganda robusta coffees submitted to different degrees of roasting was isolated and characterized, and the changes in structure and amount of galactomannans and arabinogalactans were determined and discussed in relation to the data already available for arabica coffees, obtained under the same experimental conditions. The content of arabinogalactans extracted from robusta green coffee was higher than that extracted from arabica. For roasted coffees, the amount of galactomannans extracted ranged from 0.66% to 0.92% (w/w). These values were near 50% of those obtained from the arabica coffees using the same extraction procedure. However, the amount of arabinogalactans extracted from robusta coffees (0.56-0.72%) was in the range obtained from arabica. The structures of arabinogalactans and galactomannans extracted from green and roasted coffees were not sufficiently different between robusta and arabica coffees to explain the observed differences in extraction yields for the arabinogalactans from green coffees and for the galactomannans from roasted coffees. The total polysaccharide content and the structures of the galactomannans and arabinogalactans in the two green coffee varieties investigated were also very similar. These differences in the extraction of high-molecular-weight polysaccharides between arabica and robusta roasted coffees may be related to the different susceptibility of the cell walls during the roasting process, known to result in a different porosity between arabica and robusta roasted coffees.     

In vitro and ex vivo antihydroxyl radical activity of green and roasted coffee

The specific antiradical activity against the hydroxyl radical of the water soluble components in green and dark roasted Coffea arabica and Coffea robusta coffee samples, both in vitro by the chemical deoxiribose assay and ex vivo in a biological cellular system (IMR32 cells), were determined. All the tested coffee solutions showed remarkable antiradical activity. In the deoxiribose assay, all the tested solutions showed similar inhibitory activity (IA%) against the sugar degradation (IA values ranged from 45.2 to 46.9%). In the cell cultures, the survival increase (SI%) ranged from 197.0 to 394.0% with C. robusta roasted coffee being significantly more active than the other samples. The coffee solutions underwent dialysis (3500 Da cutoff membrane) to fraction their components. In both systems, the dialysates (MW < 3500 Da) either from green or roasted coffee, showed antiradical activity, while the only retentates (MW > 3500 Da) from the roasted coffee samples were active. The preparative gel-filtration chromatography of roasted coffee C. robusta dialysate gave three fractions active in the biological system, all containing chlorogenic acid derivatives. The most active fraction was found to be that containing the 5-O-caffeoilquinic acid, which shows a linear relation dose-response ranging from 0.02 to 0.10 mM. The results show that both green and roasted coffee possess antiradical activity, that their more active component is 5-O-caffeoyl-quinic acid, and moreover that roasting process induces high MW components (later Maillard reaction products, i.e., melanoidins), also possessing antiradical activity in coffee. These results could explain the neuroprotective effects found for coffee consumption in recent epidemiological studies.     

Analysis of coffee for the presence of acrylamide by LC-MS/MS

A variety of popular instant, ground, and brewed coffees were analyzed using a modified liquid chromatography-tandem mass spectrometry (LC-MS/MS) method specifically developed for the determination of acrylamide in foods. Coffee test portions were spiked with 13C3-labeled acrylamide as an internal standard prior to their extraction and cleanup. Ground coffees (1 g) and instant coffees (0.5 g) were extracted by shaking with 9 mL of water for 20 min. Brewed coffee test portions (9 mL) were taken through the cleanup procedure without further dilution with extraction solvent. Coffee test portions were cleaned up by passing 1.5 mL first through an Oasis HLB (hydrophilic/lipophilic copolymer sorbent) solid phase extraction (SPE) cartridge and then a Bond Elut-Accucat (cation and anion exchange sorbent) SPE cartridge. The cleaned up extracts were analyzed by positive ion electrospray LC-MS/MS. The MS/MS data was used to detect, confirm, and quantitate acrylamide. The limit of quantitation of the method was 10 ng/g for ground and instant coffees and 1.0 ng/mL for brewed coffee. The levels of acrylamide ranged from 45 to 374 ng/g in unbrewed coffee grounds, from 172 to 539 ng/g in instant coffee crystals, and from 6 to 16 ng/mL in brewed coffee.     

Chemical characterization of the high molecular weight material extracted with hot water from green and roasted arabica coffee

The polysaccharides present in coffee infusions are known to contribute to the organoleptic characteristics of the drink, such as the creamy sensation perceived in the mouth known as "body", the release of aroma substances, and the stability of espresso coffee foam. To increase the knowledge about the origin, composition, and structure of the polysaccharide fraction, the high molecular weight material (HMWM) was extracted with hot water from two green and roasted ground arabica coffees: Costa Rica (wet processed) and Brazil (dry processed). The polysaccharides present in the green coffees HMWM were arabinogalactans (62%), galactomannans (24%), and glucans, and those found in roasted coffees were galactomannans (69%) and arabinogalactans (28%). The polysaccharides of the HMWM of the roasted coffees were less branched than those of the green coffees. The major green coffee proteins had molecular weights of 58 and 38 kDa, and the 58 kDa protein had two subunits, of 38 and 20 kDa, possibly linked by disulfide bonds. The protein fraction obtained from roasted coffees had only a defined band with < or =14 kDa and a diffuse band with >200 kDa. The majority of the galactomannans were precipitated with solutions of 50% ethanol, and the size-exclusion chromatography of the roasted fractions showed coelution of polysaccharides, proteins, phenolics, and brown compounds. The use of strong hydrogen and hydrophobic dissociation conditions allowed us to conclude that the phenolics and brown compounds were linked by covalent bonds to the polymeric material.     

Development of a solid-phase cleanup and portable rapid flow-through enzyme immunoassay for the detection of ochratoxin a in roasted coffee

A membrane-based flow-through enzyme immunoassay (patent application pending) for the detection of ochratoxin A (OA) in roasted coffee was developed. First, an extraction and solid-phase cleanup method was developed. A high partition coefficient for OA in the mobile phase was achieved by using methanol/5% aqueous NaHCO(3) as the sample extraction and cleanup solvent. The solid-phase (aminopropyl) cleanup was developed to chromatographically elute OA but retain cross-reacting compounds. Without using aminopropyl cleanup, cross-reacting compounds resulted in 100% false positives for both flow-through enzyme immunoassay and HPLC methods. However, after cleanup with aminopropyl, no false positives were observed. The flow-through results were visually evaluated. The sensitivity achieved for the flow-through was 4 microg kg(-1) in spiked roasted coffee. The assay was used to screen roasted coffee samples. Results were confirmed with HPLC with a detection limit of 1 microg kg(-1).     

Analysis of ochratoxin a in coffee by solid-phase cleanup and narrow-bore liquid chromatography-fluorescence detector-mass spectrometry

A method for the analysis of ochratoxin A (OTA) in green and roasted coffee has been developed. OTA was extracted from coffee with 1% NaHCO(3), and the extract was filtered and purified by solid-phase cleanup using a polymeric column that exhibits reversed-phase and anion-exchange functionalities. OTA was analyzed on a narrow-bore reversed-phase C(18) HPLC column with acetonitrile/water (0.1% formic acid) (40:60) as mobile phase and quantified with a fluorescence detector. The presence of OTA in coffee was confirmed by single-quadruple mass spectrometry using an electrospray ionization source. The method has been validated, obtaining a recovery of 82.5% and a detection limit of 0.1 ng/g. It has been applied to 20 coffee samples from various countries and different manufacturers with no detection of OTA.