Mycotoxin Research, Vol. 8 (1992)

The effect of roasting on the fate of aflatoxin 8,in artificially contaminated green coffee beans

Micco C*, Miraglia M, Brera C, Desiderio C, and Masci V

Istituto Superiore di Sanita - Laboratorio Alimenti -,Reparto Chimica dei Cereali,Viale Regina Elena, 299, 00161 Rome, Italy

* To whom reprint request and correspondence should be addressed

Abstract

A study was undertaken to evaluate aflatoxin 8, contamination in coffee beans.41 samples of green coffee were collected from large lots of material by repre­

sentative sampling. The raw samples were analyzed and showed no detectable levelsof aflatoxin 8,. In order to establish the heat stability of the toxin, 3 artificiallycontami nated samples (average level 10 J.1g / kg) were roasted at ca 200°C for differentoperation times periods so as to reproduce light and dark roasting procedures.Each sample was roasted both electrically and by gas.

The percentage of toxin destruction was up to 93 % for light roasted and 99 %for dark roasted coffee with a slightly higher rate up to 100 % for the electricallyroasted coffee for light and dark roasting. In order to evaluate the potential mi­gration of the aflatoxin 8, into the coffee beverage, 1 sample found contaminatedafter roasting treatment (O.SJ.1g/kg) was extracted using each of the 3 most commontypes of coffee makers. Additional destruction of the toxin was observed (up to99 %) in two cases while only 75 % of fate was obtained in the third.

The process from raw coffee beans to beverage showed a meaningful destructionof aflatoxin 8" ranging from 97 to 100 % depending on the extraction techniqueadopted in the preparation of the beverages.

Introduction

The occurrence of mycotoxin contamination in green coffee beans has, so far.never reached levels critical for human health. This is probably due to the proveninhibitory effects of caffeine on both the growth of toxigenic Aspergillus and Peni­cillium fungi and their production of mycotoxins such as aflatoxins, ochratoxins,sterigrnatocystin, citrinin, and patulin (1 - 5).

Nevertheless, the climatic conditions in the areas of coffee production prompta continuous control of the incidence of mycotoxin contamination. Most of thepapers relating to this problem have focused attention on commercial coffee samples,but no information is available on the contamination level in large lots of coffeebeans obtained by a statistically based sampling plan.

93

The obtaining of a laboratory sample representative of the lot from which it isdrawn is in fact crucial to the analysis of mycotoxins in food (6), in order to obtaina better evalutation of the actual toxin content and more accurate informationon the safety of imported commodities.

Representative sampling is necessary to better estimated levels of toxin contentin coffee beans and to decrease the probability to obtain false positive and / orfalse negative samples.

The aim of this study was to investigate the status of aflatoxin BI (AFB I ) con­tamination in several lots of imported green coffee beans from the major producingcountries.

The samples were collected on the basis of a statistical sampling plan as in ourprevious paper on ochratoxin A contamination of green coffee beans (7).

In addition the percentage of destruction of aflatoxin BI after the roastingprocess and the potential migration of the toxin into the beverage were also eva­luated.

Material and methods

Samples

41 green coffee samples (from Indonesia, Ivory Coast, Zaire, Cameroon, Santos,Kenya, Costa Rica, and Columbia) were collected by sampling single lots rangingin size from 50 to 4,000 bags (average weight of each bag: 70kg).

Green coffee was not flown directly from the producing countries, but wassampled during the storage period at two ports of entry into Italy (Trieste andGenova).

Sampling was carried out following EEC procedures (10) for aflatoxins in feedas a guideline, considering the lack of a specific sampling plan for the commodityunder examination. Depending on lot size, a variable number of samples (4 kg each)were selected. Each sample was then properly mixed and 1kg subsamples wereobtained by "quartering" technique.

The collected subsample (I kg) was finely ground and thoroughly blended, andanalytical samples (50 g) were taken by further quartering subdivison.

For 2 samples (Columbia) a preliminary freezing was necessary before grinding,probably because of the moisture content of the beans under examination.

Roasting procedure

The green coffee beans were artificially contaminated as follows: the beans weresoaked in a benzene -acetonitrile solution (98: 2) of AFB I and after a proper blend­ing the solvent was removed in a rotary evaporator under vacuum.

The samples used in the spiking experiments have been previously analyzed forthe presence of AFBI. No toxin was found in any of the samples tested.

3 samples of artificially contaminated samples were roasted by using two typesof laboratory roasters, ie electrical and by gas, in order to reproduce as closely aspossible the industrial roasting condition.

Coffee beverage

Beverages were prepared after roasting from an artificially contaminated sample(0.8pg/kg) employing 3 different kinds of coffee preparation commonly usedin Italy: bar, mocha, and by infusion.

94

Mycotoxin Research, Vol. 8 (1992)

Bar method

19 g of contaminated coffee powder were extracted with water (150 mL). Theyield of coffee beverage was 96.6 %.

Mocha method

16 g of the contaminated sample were extracted with water (165 mL). The yieldof coffee beverage was 43.6 %.

By infusion

7 g of the contaminated powder were extracted with water (200 mL). The yieldof beverage was 84 %.

Analytical procedure

AFB, level was determined by employing Paulsch et al method (8) with the follow­ing modifications:

The size of each analytical sample was reduced to 20 g.In the cleanup step the passage on C 18 Sep - Pak cartridge was omitted sincethe HPLC chromatogram of the eluate from Florisil cartridge showed no inter­ference in the elution zone of AFB!.

The HPLC apparatus used was a liquid chromatograph Perkin Elmer series lOLC(Pump A: water / acetonitrile / methanol, 130/ 70/40; Pump B: saturated iodinesolution) equipped with a Lichrospher 100 RP 18 column (Merck) fitted up witha precolumn C 18 Lichrospher (Merck).

Detection was performed by a Perkin Elmer LS4 Fluorescence spectrometer(wavelength of excitation 365 nm - emission 425 urn), and registration of datawas obtained by a Varian 4290 electronic integrator.

The confirmation of positive results was performed by adding known amountsof AFB, standards to the positive samples. In addition the contaminated sampleswere reinjected omitting the derivatization step with iodine: the absence of AFB(peak resulted as a confirmatory test. The detection limit attainable for AFB I was0.1J1g/kg for coffee beans and 0.01 J1g/kg for the beverage.

In order to evaluate the accuracy and precision of the method employed, 5 re­coveries were carried out both on green and on roasted coffee. The average recoveryvalue calculated for 5 determinations at spiking levels of 1 and 5 J1g/kg was 95 %and 98 %, respectively with a standard deviation of 4.5 and 3 and a coefficient ofvariation of 4.7 % and 3 %. These values were found satisfactory for the reliabilityof the modified method for the matrix under examination.

Results and discussion

The results of the analysis of coffee samples show that none of the 41 sampleswere contaminated by AFB 1 demonstrating good quality coffee lots strictly relatingto the lot under analysis.

The presence of caffeine probably acts in this commodity as a protective agentagainst AFB, production, although coffee plants grow in climatic conditions favor­able for the growth of Aspergillus moulds.

In order to evaluate the fate of AFB1 following the industrial process 3 samplesof green coffee were artificially contaminated at levels of 8, 9, and 11 J1g/ kg, re­spectively.

95

Tab 1: Effect of roasting procedure on aflatoxin 8, (Jlg/kg) in arti­ficially green coffee samples

Green coffeeRoasting

LightGas Electrical

DarkGas Electrical

Sample 1(Ivory Coast - Robusta) 0.8a (90 %)b(8Jlg/kg)

Sample 2(Santos - Arabica) 0.8 (93 %)(ll/lg/kg)

NDC (100%) 0.1 (99%) ND

0.1 (99%) 0.2 (98%) NO

0.3 (97 %) 0.2 (98 %) 0.1 (99 %)Sample 3(Columbia - Washed) 0.8 (91 %)

(9/lg/kg)BarMochaInfusion

0.04d (95%)0.20 (75%)0.01 (99 %)

(99.6%)e(97.7%)(99.9%)

a) Mean of triplicate analysisc) Less than detection limite) Total destruction (%)

b) AFB, destruction (%)

d) Calculated on powdered coffee

After evaporation of the solvent, coffee samples were roasted at 2000e for 8

to IOmin in the electrical treatment and at zro'c for 15 to 20min in the procedureperformed by gas.

By these procedures, two degrees of roasting were achieved, light and dark.The latter is more commonly employed in Italy.The results (Table 1) showed a marked destruction of AFB 1 in all of the 3 samples

at levels ranging from 90% to approximately 100%. Nevertheless a slight differencewas noted in electrical as opposed to the gas procedure.

was more consistent in the dark roasting than in thelight roasting procedure as the similar results obtained by Levi (9).

A study was also performed to evaluate the potential migration of the toxininto the beverage. 1 sample of roasted coffee (0.8 /lg/ kg) was used to prepare 3 kindsof coffee beverages: bar, mocha, and infusion.

All results obtained were calculated relating to powdered coffee.A very low level of AFB 1 (0.01 - 0.2 /lg/ kg) was detected in all 3 beverages

but a higher level of the toxin was detected in the sample obtained by the mochaprocedure.

This particular finding might be caused by a higher extraction power of themocha method.

In conclusion, the whole process from raw coffee to the beverage showed amarked destruction of AFB 1 ranging from 97 to 100% depending on the extractiontechnique employed.

This might presumably assure the safety of coffee from aflatoxin contamination,taking also in account that this particular commodity is naturally quite resistantto the production of AFB 1 because of the presence of caffeine in the matrix. Never­theless, in our opinion a control of the incidence of mycotoxin contaminationin coffee should be performed, considering both of the climate characteristic ofthe areas where coffee grows, and the shipping conditions of bags particularly favour­able to a propagation of toxigenic moulds.

96

Mycotoxin Research, Vol. 8 (1992)

Acknowledgements

The authors are grateful to AIIPA (Associazione Italiana Industrie Prodotti Alimen­tari) for its valuable technical assistance.

References

Buchanan RL and Flechter AM (1978) Methylxanthine inhibition of aflatoxinproduction. J Food Sci 43: 654 - 655

2 Nartowicz VB, Buchanan RL, and Segall S (1979) Aflatoxin production inregular and decaffeinated coffee beans. J Food Sci 44(N2):446 - 448

3 Buchanan RL, Tice G, and Marino D (1981) Caffeine inhibition of ochratoxinA production. J Food Sci 47: 319 - 321

4 Buchanan RL, Harry MA, and Gealt MA (1983) Caffeine inhibition of sterig­matocystin, citrinin, and patulin production. J Food Sci 48: 1226 - 1228

5 Buchanan RL, Hoover DG, and Jones SB (1983,b) Caffeine inhibition of afla­toxin production: Mode of action. Appl Environ Microbio146: 1193 - 1200

6 Whitaker TB (1977) Sampling granular foodstuffs for aflatoxin. Pure ApplChern 49: 1709 - 1717

7 Micco C, Grossi M, Miraglia M, and Brera C (1989) A study of the contamina­tion by ochratoxin A of green and roasted coffee beans. Food Add Cont 6(N3):333 - 339

8 Paulsch WE, Sizoo EA, and van Egmond HP (1988) Liquid chromatographicdetermination of aflatoxins in feedstuffs containing citrus pulp. J Assoc OffAnal Chern 71: 957 - 964

9 Levi C (1980) Mycotoxins in coffee. J Assoc Off Anal Chern 63: 1282 - 128510 Official Journal of the Community (The) (1976) NL102/1 -7

Manuscript received Jan 2, 1992; accepted May 22, 1992.

97