Findings of the EFSA Data Collection on Polycyclic Aromatic Hydrocarbons in Food

Transcription

1 EFSA/DATEX/002 (revision 1) A Report from the Unit of Data Collection and Exposure on a Request from the European Commission Findings of the EFSA Data Collection on Polycyclic Aromatic Hydrocarbons in Food First issued on 29 June 2007 and revised on 31 July 2008 European Food Safety Authority EFSA Switchboard Largo Natali Palli 5/A EFSA Fax I Parma

2 Table of Content EXECUTIVE SUMMARY... 3 INTRODUCTION... 5 Background... 5 Legislative framework... 9 Terms of Reference (ToR) DATA COLLECTION AND COLLATION (TOR 1) Response to call for information Sample submissions Analytical Methods Overview of PAH-levels found BaP levels in Codex food categories PAH-levels in oils and fats (excluding cocoa butter) PAH-levels in smoked meat and smoked meat products PAH-levels in smoked fish and some smoked fishery products PAH-levels in muscle meat of fish other than smoked fish...19 PAH-levels in crustaceans, cephalopods, other than smoked PAH-levels in fresh bivalve molluscs PAH-levels in foods for infants and young children Requested testing of cocoa butter Requested testing of food supplements Requested testing of dried fruits Requested testing of smoked canned fish INFLUENCE OF DIFFERENT PRODUCTION CONDITIONS (TOR 2) Smoking temperature Smoking source Smoke generation Smoke distribution Smoking time BAP AS AN INDICATOR FOR TOTAL PAH (TOR 3) Overall relationship Relationship with Benz[a]anthracene - BaA Relationship with Benzo[b]fluoranthene - BbFA Relationship with Benzo[j]fluoranthene - BjFA Relationship with Benzo[k]fluoranthene - BkFA Relationship with Benzo[ghi]perylene - BghiP Relationship with Chrysene - CHR Relationship with Cyclopenta[cd]pyrene - CPP Relationship with Dibenz[ah]anthracene - DBahA...38 Relationship with Dibenzo[ae]pyrene - DBaeP Relationship with Dibenzo[ah]pyrene - DBahP Relationship with Dibenzo[ai]pyrene - DBaiP Relationship with Dibenzo[al]pyrene - DBalP Relationship with Indeno[1,2,3-cd]pyrene - IP Relationship with 5-Methylchrysene - MCH Relationship with Benzo[c]fluorene - BcFL SOURCES AND PREVENTION OF COCOA BUTTER CONTAMINATION (TOR 4) CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES ANNEX 1 - CODEX FOOD CLASSIFICATION

3 Executive Summary Polycyclic aromatic hydrocarbons (PAH) can be formed from a variety of combustion and pyrolysis processes. Humans can be exposed to PAHs through different routes. For nonsmokers, the major route of exposure is from food with a minor contribution from inhaled air. In cigarette smokers, the contribution from smoking and food may be of similar magnitude. Food can be contaminated from environmental sources, industrial food processing and from home food preparation. A number of PAHs have been shown to be genotoxic carcinogens. In 2002, the Scientific Committee on Food (SCF) reviewed PAH toxicity (SCF, 2002). For 15 compounds it concluded that there was clear evidence for their toxicity. In view of the non-threshold effects of genotoxic substances the Committee recommended that the levels of PAH in food should be reduced to as low as reasonably achievable. In 2005, the Joint FAO/WHO Expert Committee on Food Additives (JECFA) performed a risk assessment on PAHs, basically agreed with the SCF selection, downgraded one substance from the SCF list, and nominated one further compound for observation in food (JECFA, 2005). The combined list nominated by either SCF or JECFA would thus comprise of 16 substances. Maximum levels of benzo[a]pyrene, used as a marker for PAH contamination, in a range of foodstuffs are now specified in Commission Regulation (EC) 1881/ However, there are some uncertainties in relation to the accuracy of benzo[a]pyrene as a general indicator for overall PAH contamination and the selection of food groups specified in the regulation. The Commission thus issued Recommendation 2005/108/EC 2 asking Member States to monitor the level of priority PAHs in food and requested that the European Food Safety Authority collate and evaluate the information gathered. Seventeen Member States submitted useful results from testing of 9,714 food samples belonging to 95 different Codex food categories for the presence of up to 25 different PAHs, including the 16 nominated compounds. The maximum concentration recorded for any priority PAH was 1,064 µg/kg of benzo[a]anthracene in tinned sprats in oil, while 31.8% of samples tested negative for any PAH. Benzo[a]pyrene concentrations over the detection limit were found in 72 of the 95 Codex food categories or in 47% of the samples tested. Only 36 food categories or 13.4% of the samples had concentrations exceeding 1 µg/kg and 16 food categories or 2.3% of the samples had concentrations exceeding 10 µg/kg. Samples belonging to food categories covered by Regulation (EC) 1881/2006 exceeded the respective limits for benzo[a]pyrene in up to 7.3% of cases (processing factors for preserved fish samples could not be calculated). However, some other PAHs were present at higher or much higher concentrations. Testing of some other food categories revealed relatively high concentrations of several PAHs in cocoa butter, canned smoked fish and food supplements while dried fruit had considerably lower concentrations. 1 OJ L 364, , p. 5 2 Commission recommendation of 4 February 2005 on the further investigation into the levels of polycyclic aromatic hydrocarbons in certain foods, OJ L 34, , p.43. 3

4 There was a lack of reporting of production conditions associated with the different samples. It was possible to elucidate that smoking temperature had a significant influence on the formation of PAH in that higher temperatures were associated with higher PAH levels. The assumption that benzo[a]pyrene is a good indicator of any PAH contamination was proved dubious. Of the samples tested for all of the15 SCF priority PAHs, 33% showed concentrations for one or more PAH above the limit of detection with benzo[a]pyrene concentrations below the limit of detection. Results varied across food categories and the different PAHs. Chrysene was the most problematic compound with 38% above the limit of detection and concentrations of up to 343 µg/kg found in samples with benzo[a]pyrene below the limit of detection. Benzo[c]fluorene, the compound highlighted by JECFA, had the second highest maximum of almost 27 µg/kg in a sample testing negative for benzo[a]pyrene. In view of these findings, the suitability of maintaining benzo[a]pyrene as a marker needs to be carefully assessed, alongside with other possible risk management options. The results also point to a problem with high levels of PAH found in cocoa butter, canned smoked fish and food supplements that might be considered for separate legislative action, while it seems possible to produce dried fruits without elevated levels of PAH. 4

5 Introduction Background Polycyclic aromatic hydrocarbons (PAH) form a class of diverse organic compounds, each of them containing two or more aromatic rings. Hundreds of different such compounds may be formed and released during a variety of combustion and pyrolysis processes and thus the natural and anthropogenic sources of PAHs in the environment are numerous. PAH compounds are emitted from processing of coal, crude oil, petroleum, and natural gas, from production of aluminium, iron and steel, from heating in power plants and homes (oil, gas, charcoal-fired stoves, wood stoves), burning of refuse, wood fires, and from motor vehicle exhausts (SCF, 2002). A number of PAHs have been shown to be genotoxic carcinogens (IARC, 1973, 1983, 1984, 1985, 1987, 1989; US EPA, 1984; Montizaan et al., 1989; IPCS, 1998). The best studied of the PAHs is benzo[a]pyrene and the International Agency for Research on Cancer (IARC) concluded in 1987 that it is a probable human carcinogen (IARC, 1987). Humans can be exposed to PAHs through different routes. In cigarette smokers, the contributions from smoking and food may be of a similar magnitude. For non-smoking humans, the major routes of exposure to PAH are from food and to some extent from inhaled air. Food can be contaminated by: (i) (ii) (iii) environmental PAHs that are present in air, soil or water; industrial food processing methods; and home food preparation. PAHs have been detected in a variety of foods, notably vegetables as a result of the deposition of airborne PAHs, and in fish and mussels from contaminated waters (Edwards, 1983; Nielsen et al., 1996). The waxy surface of vegetables and fruits can concentrate low molecular mass PAHs through surface adsorption and particle-bound high molecular mass PAHs can contaminate the surface due to atmospheric fallout. PAHs can also contaminate foods during industrial smoking, heating and drying processes that allow combustion products to come into direct contact with food. Contamination of cereals and of vegetable oils (including seed oils and olive residue oils) with PAH usually occurs during technological processes like direct fire drying, where combustion products may come into contact with the grain, oil seeds or the oil (Speer et al., 1990; SCFS, 2001). PAHs are also formed as a result of certain food preparation methods, such as grilling, roasting and smoking. The highest PAH concentrations are usually found in charcoal grilled/barbecued foods (especially meat and meat products grilled under prolonged and severe conditions), foods smoked by traditional techniques (fish in particular), and mussels and other seafood from polluted waters (Guillen et al., 1997; Phillips, 1999). Smoked and grilled food may contribute significantly to the intake of PAHs if such foods are a large part of the usual diet. For example, grilled/barbecued meat was the second highest contributor, after the bread, cereal and grain group, in a U.S. study (Butler et al., 1993). However, generally the major contributors to PAH intake in the average diet are oils and fats, cereals, fruits and vegetables. 5

6 The Scientific Committee on Food (SCF) reviewed the presence and toxicity of PAHs in food and issued an opinion on 4 December 2002 (SCF, 2002). For fifteen compounds (benz[a]anthracene, benzo[b]fluoranthene, benzo[j]fluoranthene, benzo[k]fluoranthene, benzo[ghi]perylene, benzo[a]pyrene, chrysene, cyclopenta[cd]pyrene, dibenz[a,h]anthracene, dibenzo[a,e]pyrene, dibenzo[a,h]pyrene, dibenzo[a,i]pyrene, dibenzo[a,l]pyrene, indeno[1,2,3-cd]pyrene and 5-methylchrysene) it concluded that there was clear evidence of mutagenicity/genotoxicity in somatic cells in experimental animals in vivo. With the exception of benzo[ghi]perylene there was also clear carcinogenic effects in various types of bioassays in experimental animals. Although only benzo[a]pyrene has been adequately tested using dietary administration, SCF stated that these compounds may be regarded as potentially genotoxic and carcinogenic to humans and represent a priority group in the assessment of the risk of long-term adverse health effects following dietary intake of PAHs. For six other compounds (anthranthene, benzo[ghi]fluoranthene, benzo[c]phenanthrene, 1- methylphenanthrene, perylene, triphenylene) the evidence of genotoxicity was limited and mainly based on results obtained in vitro. Further studies, especially in vivo, were recommended to clarify the genotoxic potential of these PAHs. Equivocal or contradictory data were available for another eight compounds (acenaphthene, acenaphthylene, benzo[b]fluorene, benzo[e]pyrene, coronene, fluoranthene, fluorene, phenanthrene), which cannot be properly evaluated for genotoxicity. Finally, four compounds (anthracene, benzo[a]fluorene, naphthalene, pyrene) gave totally or mainly negative results in a variety of short term tests. Data from EU surveys indicate that the estimated maximum dietary exposure of adults to each of the most abundant PAHs such as anthracene, phenanthrene, fluoranthene and pyrene may be in the range of ng/kg bw/day. The dietary exposure to the other PAHs, including the 15 PAHs nominated by the SCF, all but one considered to be both potentially genotoxic and carcinogenic for humans, would be one order of magnitude lower. Thus the estimated maximum daily intake of benzo[a]pyrene from food is approximately 6-8 ng/kg bw/day for a person weighing 70 kg. This estimated maximum daily intake is about 5-6 orders of magnitude lower than the daily doses observed to induce tumours in experimental animals. The SCF concluded that at these levels of intake non-carcinogenic effects are not to be expected and that the risk of heritable effects from dietary exposure to PAHs is low. However, in view of the non-threshold effects of genotoxic substances the levels of PAHs in foods should be reduced to as low as reasonably achievable. The SCF also concluded that benzo[a]pyrene may be used as a marker of occurrence and effect of the carcinogenic PAHs in food, based on examinations of PAH profiles in food and on evaluation of a recent carcinogenicity study of coal tars in mice. A conservative assessment would imply that the carcinogenic potency of total PAH in foods would be 10 times that contributed by benzo[a]pyrene alone. The Committee however stressed that though it considers benzo[a]pyrene as a marker of carcinogenic PAH in food, chemical analyses should continue to collect data on the whole PAH profile in order to be able to evaluate the contamination of food commodities and any future change in the PAH profile. In 2005, the Joint FAO/WHO Expert Committee on Food Additives (JECFA) performed a risk assessment on PAHs and estimated margins of exposure (MOE) as a basis for advice on 6

7 compounds that are both genotoxic and carcinogenic (JECFA, 2005). JECFA compared mean and high-level intakes of PAHs with the calculated benchmark dose lower confidence limit for PAHs and calculated MOEs of 25,000 and 10,000 respectively. Based on these MOEs, JECFA concluded that the estimated intakes of PAHs were of low concern for human health. The Committee recommended that future monitoring should include, but not be restricted to, analysis of 13 of the PAHs earlier identified as being both genotoxic and carcinogenic by the SCF excluding cyclopenta[cd]pyrene. JECFA also recommended that benzo[c]fluorene be included in further analyses of food to help inform future evaluations since levels of benzo[c]fluorene-derived adducts were much higher than those of benzo[a]pyrene-derived adducts in the lungs of rats fed with coal tar and JECFA found no data on its occurrence in food. This report will focus on 16 priority PAHs, the 13 PAH compounds earmarked for monitoring by both SCF and JECFA, the two additional compounds nominated by SCF and the one additional compound nominated by JECFA. There will also be brief mention of other compounds tested by Member States. Table 1 lists the substances appearing in the report and the abbreviations used. Table 1: Polycyclic aromatic hydrocarbons included in Member State results submissions and their priority classification according to SCF and JECFA. Common name Chemical structure Abbreviation SCF/JECFA PRIORITY Benz[a]anthracene BaA Benzo[b]fluoranthene BbFA Benzo[j]fluoranthene BjFA Benzo[k]fluoranthene BkFA Benzo[a]pyrene BaP Chrysene CHR Dibenz[a,h]anthracene DBahA Dibenzo[a,e]pyrene DBaeP Dibenzo[a,h]pyrene DBahP 7

8 Common name Chemical structure Abbreviation Dibenzo[a,i]pyrene DBaiP Dibenzo[a,l]pyrene DBalP Indeno[1,2,3-cd]pyrene IP 5-Methylchrysene SCF PRORITY CH 3 MCH Benzo[ghi]perylene BghiP Cyclopenta[cd]pyrene JECFA PRIORITY CPP Benzo[c]fluorene BcFL OTHER Acenaphthene AC Acenaphthylene Anthracene ACL AN Benzo[bjk]fluoranthene Sum of BbFA, BjFA and BkFA BbjkF Fluoranthene FA Fluorene Naphthalene Phenanthrene FL NA PHE Pyrene PY 8

9 Legislative framework Council Regulation (EEC) No 315/93 3 laying down Community procedures for contaminants in food provides a European framework for setting maximum levels for certain contaminants in foodstuffs. In 2001 the Commission introduced maximum levels for some contaminants through Regulation (EC) No 466/2001 4, which has now been replaced by Commission Regulation (EC) No. 1881/2006. The Scientific Committee on Food issued an opinion on polycyclic aromatic hydrocarbons in December 2002 in which it was concluded that benzo[a]pyrene could be used as a marker for the occurrence and carcinogenic effects of PAHs. On the basis of this opinion and the data collected by the Member States within the framework of Council Directive 93/5/EEC 5 on the assistance to the Commission and co-operation by the Member States in the scientific examination of questions relating to food (SCOOP Report task ), the following maximum levels for benzo[a]pyrene were laid down (Table 2). Table 2: Maximum levels (ML) of benzo[a]pyrene specified in Regulation (EC) No. 1881/2006. See the original document for the precise text as amended by further explanations. Foodstuff ML (µg/kg wet weight) Oils and fats (excluding cocoa butter) intended for direct human consumption or use as an ingredient in foods 2.0 Smoked meats and smoked meat products 5.0 Muscle meat of smoked fish and smoked fishery products, excluding bivalve molluscs. The maximum level applies to smoked crustaceans, excluding the brown meat of crab and excluding head and thorax meat of 5.0 lobster and similar large crustaceans (Nephropidae and Palinuridae). Muscle meat of fish, other than smoked fish 2.0 Crustaceans, cephalopods, other than smoked. The maximum level applies to crustaceans, excluding the brown meat of crab and excluding head and thorax meat of lobster and similar large crustaceans (Nephropidae and 5.0 Palinuridae). Bivalve molluscs 10.0 Processed cereal-based foods and baby foods for infants and young children 1.0 Infant formulae and follow-on formulae, including infant milk and follow-on milk 1.0 Dietary foods for special medical purposes intended specifically for infants 1.0 In some foods such as dried fruit, food supplements and cocoa butter benzo[a]pyrene has been found, but available data were inconclusive as to the levels that are reasonably achievable. To gather more data on levels of PAH in foodstuffs and to clarify the question whether benzo[a]pyrene could be maintained as a marker for the group of PAHs, the Commission asked Member States to monitor PAHs in Commission Recommendation 2005/108/EC. The results were reported to the European Food Safety Authority (EFSA). 3 OJ L 37, , p. 1 4 OJ L 77, , p. 1 5 OJ L 52, , p

10 Terms of Reference (ToR) In order to enable the Commission to make use of the results of the investigations carried out in the framework of Recommendation 2005/108/EC, EFSA was asked to: 1. Collect and collate data on PAHs in food in general and specifically in relation to the food categories listed in Regulation 1881/2006, as well as for some additional commodities, i.e. dried fruits, food supplements and cocoa butter, in order to determine what levels are reasonably achievable. 2. Determine the levels of PAHs found in different food categories under different production conditions (e.g. production and processing of oils and fats and different smoking, drying and heating processes of foods). 3. Further analyse the relative proportions of PAHs in foods to inform a review of the suitability of maintaining benzo[a]pyrene as a marker for the PAH group. In particular the 15 priority PAHs set out in the Annex of Recommendation 2005/108/EC should be considered. Benzo[c]fluorene 6 should be considered in addition, if possible. 4. Investigate the possible sources of contamination of cocoa butter as well as methods for prevention of contamination of benzo[a]pyrene in cocoa butter. Data Collection and Collation (ToR 1) The Terms of Reference asks for information on PAH levels in food in general and in some specified food categories. A brief overview will be provided describing the data collection process and detailed results presented of the distribution of BaP only across general food categories and for the 16 priority PAHs across 13 selected food groups of particular regulatory interest. Response to call for information In response to the call for further information on PAHs in selected foods, EFSA assembled a project steering committee with participants from Member States and the Commission. A joint SANCO/EFSA/JRC workshop was organised in July 2005 attracting 74 delegates from across the European Union. Following the workshop, a call for submissions of data on PAHs in food was issued with the invitation open from 10 October 2005 to 10 October Data was to be submitted from laboratories involved in official food control, research, and the food industry. At the end of the period a reminder was issued to all EU Member States and the call finally closed in February Initially a database for direct submission of data through a web interface was developed. However, Member States found this submission system cumbersome since all information had to be re-typed and thus an MS Excel spreadsheet was developed as an alternative submission form. The form included space for entering all the 15 genotoxic compounds identified by SCF, the substance nominated by JECFA and fields for other PAHs if required. It also contained space to enter details about analytical methodology, food description and processing conditions. 6 As recommended by JECFA (JECFA, 2005) 10

11 Sample submissions In total, 18 countries submitted results from analyses of PAHs in a variety of different food products. The submission from Hungary contained aggregated data only and even data in ranges so could not be included in the database. Valid results were received from testing of 9,714 products as illustrated in Table 3. Table 3: Number of results reported from respective country and the spread of food categories covered (major contributor is indicated when the Member State covered 50% of sample numbers and sole contributor when the Member State was the only one submitting samples for the category). Samples Food categories covered Country Number Percent Total Major contributor Sole contributor Austria Belgium Cyprus Czech Republic Denmark Estonia Finland France Germany 6, Greece Ireland Italy Latvia Slovakia Spain Sweden United Kingdom Germany submitted the most results consisting of 64.3% of the overall material. However, 2,584 of those results included only BaP values and will be used when looking at the level of contamination of individual food categories, but cannot be used when comparing relationships between the different PAHs. The products were categorised using the Codex Food Classification system (see Annex 1) into as specific food categories as possible. The product mix varied considerably between submissions from the different Member States. In total, results were reported from 95 of the 270 available food classifications with a major contribution ( 50% of samples) from one Member State of 45 and sole coverage of 42. When interpreting results it is important to recognise the possible country bias for categories with exclusive submissions from one Member State only. Some bias is also possible related to the sampling method. The samples originated from both targeted and random sampling and although an overall total for all food categories is shown at the end of Table 5, this is heavily influenced by preferential sampling of food categories covered by legislated maximum levels. 11

12 Analytical Methods Official analytical methods are available for some of the relevant PAHs in certain food matrices, but in general there is a lack of officially approved methods and in particular they do not include many of the dibenzopyrenes or 5-methylcrysene. Other methods have been used to quantify the latter groups, which should be kept in mind when considering the results reported. Following homogenization of the foodstuff, PAHs are extracted using different techniques prior to clean up and measurement. PAHs are most often identified and quantified using either gas chromatography (GC) with flame ionization detection (FID) or coupled to mass spectrometry (MS) or high performance liquid chromatography (HPLC) with ultraviolet (UV) or fluorescence (FL) detection or coupled to MS. The application of FID is not anymore state of the art. It lacks selectivity and does not allow the use of isotope labelled internal standards. In response to a question of the analytical method used 4% indicated that they used GC-FID, 28% that they used HPLC-FL, 26% that they used HPLC-UV/FL and 43% that they used GC-MS. In addition to potential losses of PAH during homogenization, extraction and clean up, there are a number of other factors that may lead to erroneous results. During sample collection and storage it is important that the sample not be exposed to tobacco smoke, light and high temperatures (leading to volatilization and/or chemical conversion). Also storage for a prolonged time before analysis may result in the reaction of some PAHs with components of the food matrix. Details on sample preparation were given in submissions, the most common methods being liquid/liquid extraction and saponification in a shaking water bath at 60 C. Attention should also be paid to the possible co-elution of some PAHs. For example, under the gas chromatographic conditions generally used, chrysene + triphenylene, the benzo[b+j+k]fluoranthenes, and the dibenzo[a,h+a,c]anthracenes may co-elute and give rise to only a single peak. When HPLC is used the separation of benzo[b]fluoranthene + perylene, benzo[k]fluoranthene + dibenz[a,c]anthracene, and benzo[j]fluoranthene + benzo[e]pyrene may be critical. Descriptive statistics included the calculation of the arithmetic mean, the median, the 5 th, 90 th and 95 th percentiles (assigned as P05, P90 and P95 in the following tables), and the maximum. The accuracy of the calculations diminishes with low sample numbers (Kroes et al., 2002), but for practical purposes percentiles were calculated when sample numbers exceed 10. The middle bound was used when calculating descriptive statistics, that is for values below the limit of detection (LOD) a value of half the LOD was entered. In assessing concentrations found for the PAHs it is important to have a clear understanding of the sensitivity of the methods used. Some laboratories, but not all, report values as less than the LOD for samples with no detectable levels and as less than the LOQ for samples with nonquantifiable traces. Other laboratories consistently report either the LOD or the LOQ for any such case. LOD is most often defined as three times the standard deviation of a blank or a low concentration sample while LOQ is ten times the standard deviation or 3.3 times the LOD. To standardise the material the LOQ was divided by the factor of 3.3 to be entered as 12

13 the LOD for all samples with non-detected or non-quantifiable amounts. Reported or calculated LODs varied between and 1 µg/kg for the respective PAH except for cyclopenta[cd]pyrene with a maximum of 6 µg/kg (38 samples were initially excluded in full because the LOD for either benz[a]anthracene, chrysene or indeno[1,2,3-cd]pyrene was reported above 1 µg/kg and also 322 individual results for cyclopenta[cd]pyrene with a reported LOD above 6 µg/kg were marked as missing only for this compound to not unduly influence the exposure assessment). Table 4: Summary of LOD reported for the 25 PAHs. The shades of grey indicate the SCF/JECFA joint priority substances, the extra SCF substances, the added JECFA substance, and other nonpriority substances from light to dark, respectively. PAH LOD reported (μg/kg) Minimum Median Mean Maximum BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL PY PHE ACL FA AC NA FL AN BbjkF Overview of PAH-levels found Analytical results reported for individual samples included from one to 25 different PAHs. Only 1,375 of the food products were tested for the full range of the 15 priority PAHs as nominated by SCF, of which 873 also included testing of benzo[c]fluoranthene as nominated by JECFA. Only between 15 and 369 of the samples were tested for other PAHs and because of the few results they were not included in any further analysis. An overview of the results is presented in Table 5, however, the values should not be compared across the different PAHs since they represent different ranges of food categories. 13

14 Table 5: Presentation of descriptive statistics for the concentration (µg/kg) of up to 16 PAHs in 9,714 food products. PAH N >LOD Concentration in µg/kg P05 Median Mean P90 P95 Maximum BaA % BaP % BbFA % BkFA % CHR % DBahA % IP % BjFA % DBaeP % DBahP % DBaiP % DBalP % MCH % BghiP % CPP % BcFL % Of the 9,714 samples tested for the presence of one or more of the priority PAHs, 3,086 samples (31.8%) had no result for any PAH above the LOD. For only one of the PAHs, chrysene, did the proportion of results above the LOD exceed 50% and as a consequence the median reflects in most cases only the middle bound of the LOD. Among the 15 priority PAHs and benzo[c]fluorene the mean varied between 0.08 µg/kg for dibenz[a,h]pyrene to 3.23 µg/kg for chrysene. A maximum result of 1,064 µg/kg was recorded for benz[a]anthracene in a tinned sprats in oil product and a second highest result of 690 µg/kg was recorded for benzo[b]fluoranthene in a food supplement. The result for the sprats in oil product could be an aberration since no other PAH in this sample recorded a gh concentration, however, the food supplement also recorded the maximum of 590 µg/kg for chrysene, and the highest concentration also for several other PAHs. No other products came within half of the high values recorded for these two products. Benzo[a]pyrene levels in Codex food categories The level of contamination of foodstuffs with benzo[a]pyrene across 95 Codex food categories (see Annex 1) was examined (Table 6). The proportion of product exceeding set limits (mirroring maximum levels) was calculated for all food categories irrespective of legislative coverage. Samples were classified at the most disaggregated level possible and only samples that did not fit into any subcategory were placed at the uppermost aggregated level. 14

15 Table 6: Test results for benzo[a]pyrene (BaP) across 95 Codex food categories (abbreviated names used) and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. Greyed rows have results above the LOD exceeding 1 µg/kg. Codex Food Category N Product (%) over limit (μg/kg) Concentration μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max Dairy drinks Cream Cheese Unripe cheese Ripe cheese Process cheese Flavour cheese Dairy desserts Fats and oils Vegetable oils Animal fats Butter Margarine Emulsion <80% Oil-in-water Fresh fruit Untreated fruit Processed fruit Dried fruit Canned fruit Vegetables etc Fresh vegetables Untreated veg Processed veg Dried veg Veg. in oil etc Canned veg Veg not Fermented veg Cocoa products Cocoa mixes etc Cocoa, chocolate Grain whole, flakes Flours, starches Flours Breakfast cereals Pastas, noodles Processed rice Breads and rolls Yeast leavened Other bakery Cakes, cookies Fresh meat etc Processed meat Non-heated Cured Cured, dried Fermented

16 Codex Food Category N Product (%) over limit (μg/kg) Concentration μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max Heat-treated Frozen Comminuted meat Non-heated Cured Cured, dried Fermented Heated Fresh fish etc Fresh fish Fresh mollusc Processed fish etc Cooked mollusc Smoked, dried, fermented, salted Semi-preserved Marinated fish Pickled fish etc Fully preserved fish Salt substitutes Herbs, spices etc Herbs, spices Vinegars Soups and broths Ready-to-eat soup Sauces and like Emulsified sauces Non-emulsified Mixes for sauces Fresh bean curd Dried bean curd Infant formulae etc Infant formulae Follow-up form Medical formula Other food children Food supplements Waters Mineral waters Waters, soda Coffee, tea etc Alcohol beverages Beer and malt Grape wines Spirits > 15% Snacks potato etc Snacks fish based Composite foods TOTAL

17 BaP levels over the detection limit were found in foods from 71 of the 95 Codex categories or in 47% of the samples tested. The 95 th percentile concentration was 3.6 µg/kg. Only 36 food categories, or 13.4% of the samples, had levels exceeding 1 µg/kg with 219 samples (2.3%) exceeding 10 µg/kg. There were 84 mainly preserved fish or seafood samples exceeding 10 µg/kg, 43 vegetable oil samples, 31 meat samples, 22 coffee and tea samples, 18 food supplements, 12 samples of herbs and spices and 9 mussel samples (of which some had been harvested in an industrial area). PAH-levels in oils and fats (excluding cocoa butter) Overall, 2,063 oils and fats results were reported from 12 Member States. Regulation (EC) 1881/2006 specifies a maximum level of 2.0 µg/kg for BaP in oils and fats (excluding cocoa butter) intended for direct human consumption or use as an ingredient in foods. Although most samples conformed to the maximum level for BaP in oils and fats, 7.3% were in breach of this level. The maximum concentration was recorded for CHR. Findings of the 15 priority PAHs and BcFL are shown in Table 7. Table 7: PAH in oils and fats and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. Boxed area indicates the proportion of samples above the maximum level. PAH N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL PAH-levels in smoked meat and smoked meat products Overall, 1,584 smoked meat and smoked meat products results were submitted by 14 Member States. Regulation (EC) 1881/2006 specifies a maximum level of 5.0 µg/kg for BaP in smoked meat and meat products. Most samples conformed to this maximum level, with only 2.8% in breach. Findings of the 15 priority PAHs and BcFL are shown in Table 8. 17

18 Table 8: PAH in smoked meat and meat products and percentage of product exceeding the LOD, 1, 2, 5, or 10 µg/kg. Boxed area indicates the proportion of samples above the maximum level. PAH N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL PAH-levels in smoked fish and some smoked fishery products Overall, 1,060 clearly marked smoked fish and fishery product results were received from 15 Member States. Regulation (EC) 1881/2006 specifies a maximum level of 5.0 µg/kg for BaP in muscle meat of smoked fish and fishery products, excluding bivalve molluscs. Only 2.7% of the samples exceeded this level. Some high results were noted with a maximum of 200 µg/kg for CHR. Findings of the 15 priority PAHs and BcFL are shown in Table 9. Table 9: PAH in smoked fish and fishery products and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. Boxed area indicates the proportion of samples above the maximum level. PAH Count Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL

19 PAH-levels in muscle meat of fish other than smoked fish Regulation (EC) 1881/2006 specifies a maximum level of 2.0 µg/kg in muscle meat of fish other than smoked fish. The maximum level set in the legislation applies to fresh fish. For processed or composite products special processing factors have to be taken into account, information that was not supplied with the results. Thus initially muscle meat of fresh fish samples are presented in Table 10 followed by muscle meat of all fish other than smoked, including processed and composite product, in Table 11. Table 10: PAH in fresh muscle meat of fish other than smoked and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. Boxed area indicates the proportion of samples above the maximum level. PAH N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL None of the samples in the fresh fish group presented in Table 10 exceeded the maximum limit. However, the situation is different for fresh and processed fish presented in Table 11. Because the Codex system used by Member States to classify products has two distinct categories, one for smoked fish (09.2.5) and one for canned fish (09.4), but none for smoked and canned, it is likely that some canned product in Table 11 could have been smoked as well. This caveat should be noted when reviewing information in Table 11, as well as the absence of application of processing factors. Thus the high values recorded for several of the PAHs should be interpreted with caution. 19

20 Table 11: PAH in muscle meat of fish other than smoked and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. Boxed area indicates the proportion of samples above the maximum level. N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA BghiP CHR CPP DBahA DBaeP DBahP DBaiP DBalP IP MCH BcFL PAH-levels in crustaceans, cephalopods, other than smoked Regulation (EC) 1881/2006 specifies a maximum level of 5.0 µg/kg for BaP in crustaceans and cephalopods other than smoked. Overall, only 12 results from testing of samples from this category were submitted from three Member States. No sample result above the LOD was found for BaP, only one sample result each for BbFA, CHR and BghiP were found to be just above their respective LOD. PAH-levels in fresh bivalve molluscs Overall, 382 fresh bivalve mollusc results were submitted by six Member States. Regulation (EC) 1881/2006 specifies a maximum level of 10.0 µgkg for BaP in fresh bivalve molluscs. Only 2.4% of the samples exceeded this level. Findings of the 15 priority PAHs and BcFL are shown in Table

21 Table 12: PAH in fresh bivalve molluscs and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. Boxed area indicates the proportion of samples above the maximum level. PAH N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL Bivalve molluscs had a bimodal distribution for PAH in that some showed quite high concentrations while others had low concentrations. CHR, BaA and BbFA all had concentrations over 200 µg/kg. Some of the bivalve molluscs were sampled in industrially polluted areas, which would have influenced the result and could explain the bimodal distribution. PAH-levels in foods for infants and young children Regulation (EC) 1881/2006 specifies a maximum level of 1.0 µg/kg for BaP in processed cereal-based foods and baby foods for infants and young children, in infant formulae and follow-on formulae, including infant milk and follow-on milk, and in dietary foods for special medical purposes intended specifically for infants. Overall for the three groups only 102 results were submitted from eight Member States. Findings of the BaP only are shown in Table 13 since the few analytical results made a detailed analysis meaningless. Table 13: PAH in food categories for infants and young children and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. Boxed area indicates the proportion of samples above the maximum level. PAH N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max Cereal-based and baby foods for infants and young children BaP Infant and follow-on formulae BaP Special purpose medical foods for infants BaP

22 No sample result had a value above 1 µg/kg and only 3% and 2%, respectively, for cerealbased and baby foods for infants and young children, and infant and follow-on formulae, exceeded the LOD. No special purpose medical food for infants had any detectable PAH. Requested testing of cocoa butter Cocoa butter is currently exempted from the maximum level for oils and fats set by Regulation (EC) 1881/2006. According to Commission Recommendation 2005/108/EC information is required on levels of BaP and other PAHs in cocoa butter, on the sources of possible contamination as well as on possible ways to reduce the contamination. Overall, 143 cocoa butter results were submitted from testing in three countries, Spain, Germany and France (Table 14). Cocoa butter was tested for a minimum of six PAHs and a maximum of 16 PAHs, including all priority substances. BcFL had the highest mean concentration at 6.93 µg/kg followed by CHR at 3.74 µg/kg. BaP, the indicator substance for the PAHs, had a mean of 1.93 µg/kg with 95.8% of the samples showing detectable concentrations of the substance. The highest concentration recorded was 46.1 µg/kg for BbFA. Table 14: PAH in cocoa butter and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL Cocoa butter seems to be consistently contaminated with a range of PAHs judging from the 143 samples analysed. However, because of the few samples tested in only three countries caution must be taken in interpreting the results. It is not possible to comment on sources of contamination or ways of reducing contamination based on the information supplied. Requested testing of food supplements Food supplements are not currently covered by Regulation (EC) 1881/2006. In Recommendation 2005/108/EC Member States were asked to investigate also the levels of PAH in food not currently covered by the Regulation, but that are suspected to contain high levels, e.g. food supplements. 22

23 Overall, 325 food supplement results were submitted from testing in five countries, Belgium, the Czech Republic, Germany, Ireland and the United Kingdom (Table 15). BaP, the indicator substance for the PAHs, had a mean of 2.54 µg/kg with 46% of the samples showing detectable concentrations of the compound. The highest priority substance was CHR with a mean of 9.11 µg/kg and a maximum of 590 µg/kg. BbFA had an even higher maximum of 690 µg/kg. Table 15: PAH in food supplements and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. PAH N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL The product range for food supplements is large, which is partly reflected in the variation seen in PAH-values and the skewed distribution with some very high values. A total of 210 different products or product variations were analysed so most often results from testing of just one sample of each product were submitted. The maximum combined PAH concentrations were recorded in a product named Arkopharma thé noir sourced in Belgium. The second highest was found in a product named Spirulina sourced in Ireland and the third highest in an alcohol free Propolis Extract sourced from the UK. Requested testing of dried fruits As for food supplements, dried fruits are not currently covered by Regulation (EC) 1881/2006. In order to obtain information on levels of PAH, Member States were asked to collect data. Overall, 264 dried fruit results were submitted from testing in eight countries, Belgium, the Czech Republic, Estonia, France, Germany, Greece, Ireland and the United Kingdom. Results are shown in Table

24 Table 16: PAH in dried fruit and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. PAH N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL BcFL, which JECFA wanted under observation, had the highest mean and maximum concentrations at 5.11 µg/kg and 144 µg/kg, respectively. BaP had a mean concentration of only 0.10 µg/kg with 23% of the samples showing detectable levels of the substance. The highest mean on the SCF list was recorded for CHR at 0.91 µg/kg and the highest maximum for BaA at 24 µg/kg. Requested testing of smoked canned fish On the basis of a request from one Member State to set a specific maximum level for BaP in smoked canned fish, the Commission asked EFSA to assess the data collected for this food group in addition to the food commodities set out in Recommendation 2005/108/EC. It was not always clear what samples were both smoked and canned. Sardines and sprats are often smoked before canning, but can sometimes be only salted. For the purpose of this survey all canned sardines and sprats results were included in the smoked and canned group (722 samples) together with a few other samples defined as both smoked and canned. Overall, 732 smoked canned fish results fitting this description were submitted from testing in seven countries, Belgium, Estonia, Finland, Germany, Latvia, Slovakia and the United Kingdom. There was little overlap (33 samples) with the previously tested clearly marked smoked fish grouping. The distribution of PAH concentrations are shown in Table 17. BaP had a mean concentration of 2.97 µg/kg with 66% of the samples showing detectable levels of the substance. BcFL, which JECFA wanted under observation, had the highest mean concentration at 11.4 µg/kg followed by BaA, CHR and CPP. Only Estonia and the United Kingdom reported BcFL results. BcFL had a much larger proportion of high values compared to the other PAHs, but only few samples analysed. Overall, higher PAH concentrations were recorded in this grouping compared to the previously tested smoked fish group. 24

25 Table 17: PAH in canned smoked fish samples and percentage of product exceeding the LOD, 1, 2, 5, or 10 μg/kg. PAH N Product (%) above limits (μg/kg) Concentration in μg/kg >LOD >1 >2 >5 >10 Median Mean P90 P95 Max BaA BaP BbFA BjFA BkFA CHR DBahA DBaeP DBahP DBaiP DBalP IP MCH BghiP CPP BcFL Influence of different production conditions (ToR 2) A number of questions were asked in relation to food processing methods of relevance to the formation of PAHs. Unfortunately only a few of the submissions contained such information. In the following a brief overview of the results are presented despite the low sample numbers. Smoking temperature For smoked foods, details of smoking procedures were given for 579 products. Cold smoking (<35 C) was applied to 30% of products, warm smoking (35-50 C) to 11% and hot smoking (>50 C) to 59% of products (Table 18). There was a tendency for an increased concentration of BaP with increasing temperatures. Table 18: The influence of smoking temperature on BaP concentrations. Smoking method N BaP concentration (µg/kg) Median Mean P 95 Maximum Cold smoke (< 35 C) Warm smoke (35-50 C) Hot smoke (> 50 C) Smoking source The smoke generation source was given for 578 samples. The most common wood used for smoke generation was beech followed by oak and alder (Table 19). 25

26 Table 19: The influence of wood or other smoke generation source on BaP concentrations. Smoking method Valid N BaP concentration (µg/kg) Median Mean P 95 Maximum Alder Apple Beech Cherry Coniferous tree Hickory Liquid smoke Mango tree Maple Mix Oak Peat Sawdust Liquid smoke gave rise to little BaP in the product, although the number of samples was small so no firm conclusion can be drawn. Use of alder seemed to result in higher levels of BaP than other wood sources. However, that could be a combined interaction between smoke generation source, smoke temperature and smoking time applied in different combinations for different products. A detailed analysis was not possible because of the limited number of samples. Smoke generation The smoke generation method was indicated for 507 samples. Smoke generation was most commonly achieved by the burning of sawdust followed by burning of wood or woodchips (Table 20). There were a range of other methods used for individual samples like friction of wood, flaming, natural gas ovens and electrical ovens. Table 20: The influence of smoke generation method on BaP concentrations. Smoke generation N BaP concentration (µg/kg) Median Mean P95 Maximum Burning of saw dust Burning of wood or woodchips Flaming Friction of wood Humid wood shaving and sawdust Industrially smoked in an oven using a t Liquid smoke Liquid smoke sprayed Oven with natural gas, additional beech Overheated steam

27 Overheated steam seemed to create the highest concentrations of BaP in treated products, however, again caution should be applied because of the limited number of samples in the material for most treatments. Smoke distribution The smoke distribution method was given for 587 samples (Table 21). Direct smoke was used for 43% of the products, indirect for 55%, with the balance being smoke flavour used in 2% of the cases. For the indirect smoking process, 7% used unfiltered and 63% used filtered smoke with the rest unspecified. Table 21: The influence of smoke distribution source on BaP concentrations. Smoke distribution N BaP concentration (µg/kg) Median Mean Percentile 95 Maximum Direct unfiltered filtered Indirect unfiltered filtered Liquid smoke For the subset of respective direct or indirect smoke distribution where it was indicated whether or not the smoke had been filtered, there was as expected less BaP in product where the smoke had been filtered before application. Smoking time Smoking time was registered for 585 samples. The smoking time varied between 15 minutes to 15 days. Some products were salted and dried before smoking using a range of different methods. The regression of smoking time against BaP concentration in the products is shown in Figure 1. There was almost no correlation between smoking time in isolation and the amount of BaP in the product. The goodness of fit of the regression line (R 2 ) was only indicating that much less than 1% of the BaP concentrations found could be explained by the smoking time. This would partly be due to the combination of smoking times and smoking temperatures. Cold smoking would normally be applied for a much longer time than hot smoking. 27

28 Figure 1: Plot of BaP in relation to duration of smoking (values above 50 hours are not shown). BaP as an indicator for total PAH (ToR 3) The terms of reference asks for a further evaluation of the relative proportions of PAHs in food to get an indication of the suitability of maintaining BaP as a marker for the PAH group. Overall relationship In relation to the use of BaP as an indicator for the 15 PAHs identified by the SCF and the extra safety factor of 10 recommended by the SCF to cater for possible variations in levels and carcinogenicity, it is uncertain if this relates to the combined level for the group or for individual PAHs. Thus comparisons will be made for both the combined group and for individual PAHs. The level of BaP alone was compared with the combined level of all 15 PAHs including BaP in 1,375 samples tested for all PAH15. Of those samples, 164 had all results below the LOD, while 456 recorded results above the LOD for one or more PAHs with BaP below the LOD. The histogram in Figure 2 shows the ratio of PAH15 over BaP for the remaining 755 samples where both recorded values above the LOD. Assuming a factor of 10 as indicated above, all values under 10 on the X-axes comply, that is overall PAH levels are less than 10 times the BaP level. Only 296 samples analysed for all 15 PAHs complied with this rule or, if including also samples with all results below the LOD as compliant, a total of 33.5%. This calculation does not factor in differences in carcinogenicity levels. SCF did not favour the approach used for dioxin compounds calculating toxic equivalency factors for the individual substances, but such comparisons have been published (Petry et al., 1996; Yang et al., 2006). 28