Probabilistic dietary exposure to phycotoxins in a recreational ... - Nature

2 downloads 0 Views 323KB Size Report
Jul 4, 2012 - shellfish consumption and contamination by major toxins (domoic acid (DA) group, okadaic acid (OA) group and spirolides. (SPXs)).
Journal of Exposure Science and Environmental Epidemiology (2013) 23, 435 -- 441 & 2013 Nature America, Inc. All rights reserved 1559-0631/13

www.nature.com/jes

ORIGINAL ARTICLE

Probabilistic dietary exposure to phycotoxins in a recreational shellfish harvester subpopulation (France) Cyndie Picot1, Gwendolina Limon2, Gae¨l Durand2, Dominique Parent-Massin1 and Alain-Claude Roudot1 Phycotoxins, secondary phytoplankton metabolites, are considered as an important food safety issue because their accumulation by shellfish may render them unfit for human consumption. However, the likely intakes of phycotoxins via shellfish consumption are almost unknown because both contamination and consumption data are very scarce. Thus, two 1-year surveys were conducted (through the same population: recreational shellfish harvesters and from the same geographical area) to assess: shellfish consumption and contamination by major toxins (domoic acid (DA) group, okadaic acid (OA) group and spirolides (SPXs)). Recreational shellfish harvesters had been targeted as an at-risk subpopulation because they consume more shellfish than general population and because they eat not only commercial shellfish species controlled by official authorities but also their own harvests of shellfish species may be in non-controlled areas and more over shellfish species non-considered in the official control species. Then, these two kinds of data were combined with deterministic and probabilistic approaches for both acute and chronic exposures, on considering the impact of shellfish species and cooking on phycotoxin levels. For acute risk, monitoring programs seem to be adequate for DAs, whereas OAs could be a matter of concern for high consumers (their acute intakes were up to ninefold the acute reference dose (ARfD)). About chronic risk, OAs are a matter of concern. The daily OAs intakes were close to the ARfD, which is, by definition, greater than the tolerable daily intake. Moreover, SPX contamination is low but regular, no (sub)chronic SPX toxicity data exist; but in case of (sub)chronic toxicity, SPX exposure should be considered. Journal of Exposure Science and Environmental Epidemiology (2013) 23, 435--441; doi:10.1038/jes.2012.44; published online 4 July 2012 Keywords: phycotoxins; shellfish; exposure assessment; probabilistic; deterministic

INTRODUCTION Phycotoxins are ubiquitous secondary metabolites produced by toxic phytoplankton. Bivalve molluscs filter-feed on these microalgae, accumulate toxins and may be consumed by humans.1,2 Case reports have demonstrated that human dietary exposure to phycotoxins may lead to serious intoxications and death in cases of potent toxins, extreme exposure or sensitivity. However, quantitative assessments of exposure to phycotoxins aimed at evaluating the likely intake of chemicals via food3 are almost non-existent.4--7 Exposure assessment must first concern at-risk subpopulation. Recreational shellfish harvesters appear to be an at-risk subpopulation because they consume a larger quantity of seafood than the general population, because their practice is both recreational and a free source of food.8--11 They also appear to be more likely to eat contaminated shellfish because they eat not only commercial shellfish but also their own harvests of shellfish species, which are may be in non-controlled areas. Indeed, in France, the official control is performed mainly in recorded professional/industrial areas and that shellfish species are considered in the sampling plans of official monitoring programs. Moreover, in case of harvesting bans for their harvesting zones, recreational harvesters can transgress the bans and continue to harvest shellfish.12 This lack of phycotoxin-exposure assessment can be due to: (i) all around the world, the national surveillance plans are defined to prevent acute intoxications: phycotoxins are mainly analysed in

shellfish only in case of phytoplankton bloom; (ii) the quantities of seafood consumed by recreational shellfish harvesters are scarce. Therefore, it is critical to assess the phycotoxin dietary intakes from shellfish consumption in this at-risk subpopulation. To estimate the probability and severity of an adverse effect, scientific data are analysed through application of an exposure assessment methodology. It is applied to model the phycotoxin exposure of humans because of the consumption of shellfish contaminated by phycotoxins. For that, the estimated individual phycotoxin concentrations in shellfish have to be combined with available consumption data.3,13 In this study, the contamination data under analysis in the assessment were about three phycotoxins of interest to humans: domoic acid and analogues (DAs), okadaic acid and analogues (OAs) and spirolides (SPXs). Succinctly, these toxins had been chosen because of their high occurrence in the area of interest, their toxicity and their different profile of toxicity and contamination. The two historical toxins (DAs and OAs) are regulated, have caused intoxications and are regularly detected at high level in the area of interest, thus they have to be under consideration in the present study. On the contrary, SPXs are not (yet) under regulation, but they have been chosen because of the recent occurrence of such toxins in France (Atlantic coast) and as they are considered as ‘‘emerging’’ toxins more data could be needed in France. At the opposite, paralytic toxins (saxitoxin and analogues) have not been taking into account (in spite of the fact

1 Laboratoire de Toxicologie Alimentaire et Cellulaire (EA 3880), Universite´ Europe´enne de Bretagne - Universite´ de Bretagne Occidentale (UEB-UBO), 6 Avenue Victor Le Gorgeu--CS93837, 29238 Brest, France; 2IDHESA Bretagne Oce´ane - Technopoˆle de Brest-Iroise, BP 52---120 Avenue de Rochon, 29280 Plouzane´, France. Correspondence to: Professor Alain-Claude Roudot, Universite´ de Bretagne Occidentale, UFR Sciences et Techniques, Laboratoire de Toxicologie Alimentaire et Cellulaire (EA 3880), 6 Avenue Victor Le Gorgeu---CS93837, 29238 Brest, France. Tel.: þ 33 2 98 01 79 82. Fax: þ 33 2 98 01 79 80. E-mail: [email protected] Received 16 May 2011; accepted 29 March 2012; published online 4 July 2012

Probabilistic exposure assessment to phycotoxins Picot et al

436

that they are regulated and they can be quantified) because such toxins have not been reported to the area of interest for a long time (in France, such toxins are currently only detected in Mediterranean Sea, in the South). The acute syndrome called Amnesic Shellfish Poisoning is caused by DAs: gastrointestinal symptoms (including vomiting, diarrhoea or abdominal cramps) appear within 24 h after the ingestion of contaminated shellfish. Then, in more severe cases, within the next hours neurological symptoms appear (confusion, disorientation, loss of short-term memory, coma and death).14 OAs are responsible for Diarrhoeic Shellfish Poisoning with its associated symptoms (gastrointestinal distress, diarrhoea, nausea, vomiting and abdominal pain); moreover, in case of chronic exposure, they may promote tumour (in vivo study).4 SPXs are Fast Acting Toxins that cause rapid death on intraperitoneal injection into mice, but no evidence of link between SPXs and poisoning events in humans has been reported.6 Getting base levels of these phycotoxins in the most concerned shellfish, that is, bivalve molluscs is a prerequisite to any phycotoxin-exposure assessment. Such data are missing because shellfish contamination is only analysed in case of phytoplankton bloom. These considerations led us to monthly monitor these phycotoxins in such species and to conduct a 1-year survey of shellfish consumption by a population of heavy consumers from the same area, that is, recreational shellfish harvesters. Achieving a meaningful exposure assessment on phycotoxins from shellfish consumption is extremely difficult because of: (i) the lack of data relative to both consumption and contamination and (ii) temporal and geographical variations in consumption and contamination. In this study, phycotoxin dietary intakes from shellfish consumption were assessed from contamination- and consumption-data collected in the same area and relative to the same subpopulation. Our investigations were aimed at: (i) carrying out a deterministic exposure assessment to decide about the need for more sophisticated estimations, (ii) making a probabilisticexposure assessment whenever it was needed, and (iii) comparing dietary phycotoxin intakes with toxicological reference doses to determine whether phycotoxin exposure is a matter of concern for human health or not. Acute and chronic exposures were assessed in both approaches.

Brittany, France). Their shellfish consumption was investigated from February 2008 to February 2009 through two complementary methods: a Food Frequency Questionnaire (FFQ) and a food diary. The FFQ was conducted through face-to-face interviews at the harvesting sites. Finally, 512 harvesters have completed the FFQ (refusal rate equals to 4%). The average age was 57 years old (ages ranged from 19--83) and almost half of the harvesters were 60 or older (48.2%) and retired (56.5%). As this tool provides long-term consumption data, but relies upon memory, this drawback was counteracted by using the records versus time of each shellfish meal (with quantities) kept in the food diary. Moreover, this diary gave additional information such as the origin of consumed shellfish (harvest, shop, restaurant), consumption by different household members and the way shellfish had been prepared. Data were validated for bivalve and gastropod groups (for more details, see Picot et al.11). The consumption data about five of the most consumed bivalve species in the area of interest are expressed as follows in Table 1: portion sizes and daily shellfish consumption rates (both with the mean and 95th percentile (P95) as well as the raw consumption in percentage).

Shellfish contamination data. To counteract the lack of databases about phycotoxin base levels in shellfish, samples were harvested monthly, from June 2009 to June 2010, on beaches of Finiste`re selected from three criteria: (i) the presence of several bivalve species, (ii) regular shellfish harvesters and (iii) regular phycotoxin events. The analyses were made on only two among the five bivalve species, which had been previously identified as being either the most consumed species or the most contaminated from a consumption survey and a test about inter-species variability (Picot et al.15) OAs and SPXs were analysed after methanolic extraction from samples, purification by solid phase extraction and quantification by HPLC-MS/MS (high performance liquid chromatographytandem mass spectrometry), following the procedure published by Gerssen et al.16 (2009), with some modifications. DAs were extracted with a water/ methanol mixture (50/50; v/v) and analysed by HPLC-MS/MS (Picot et al.15). Contamination data are often left censored because of the limits of detection (LOD) and quantification (LOQ) of analytical methods. The GEMS/ Food--Euro framework proposed different treatments according to the prevalence of censored-data17:  The number of censored data is less than or equal to 60% and then the censored data are replaced by the corresponding LOD or LOQ divided by 2 (T1).  The number of censored data is greater than 60% and then either:

MATERIALS AND METHODS Input data Shellfish consumption data. The population of interest was a group of

*

recreational shellfish harvesters set along the coasts of Finiste`re (Western

Table 1.

*

the censored data are replaced by zero (T2a) or they are replaced by the corresponding LOD or LOQ (T2b).

Descriptive statistics of shellfish consumption data according to shellfish species and origin (harvest or purchase).

Consumption derived only from harvested bivalve

Bivalve species

Oyster Mussel Cockle Carpet shell clam Razor clam King scallop

% Of consumers

20.3 27.3 63.7 74.6 23.4 0.39

Portion size (g/portion)

Daily consumption rate (g/day)

Mean

P95

Mean

P95

36.6 69.4 109.0 73.7 27.6 0.13

172.8 396.0 271.5 259.5 167.8 NC

1.68 1.66 3.15 2.43 0.57 0.0012

7.58 10.1 13.8 10.7 3.31 NC

Consumption derived only from purchased bivalve

% Of consumers

27.2 33.6 2.60 4.70 0.00 20.3

Portion size (g/portion)

Daily consumption rate (g/day)

Mean

P95

Mean

P95

34.4 80.0 2.70 3.20 0.00 15.2

102.4 264.0 NC NC 0.00 85.5

1.89 3.77 0.07 0.11 0.00 0.45

8.12 11.6 0.26 0.47 0.00 2.50

% Of raw consumption

97.8 0.00 2.38 31.7 0.00 3.69

Consumption rates derived from the total population, including non-consumers. NC: not calculable because of an insufficient number of consumers.

Journal of Exposure Science and Environmental Epidemiology (2013), 435 -- 441

& 2013 Nature America, Inc.

Probabilistic exposure assessment to phycotoxins Picot et al

437 Table 2.

Descriptive statistics of domoic acid, okadaic acid and spirolide contamination data set by shellfish species (ng/g) according to the censored data treatment.

Toxin

Das Das OA OA SPX SPX

Bivalve species

Analyses (n)

Oysters Cockles Oysters Mussels Oysters Mussels

oLOD (%)

13 13 13 13 13 13

69.2 69.2 76.9 61.5 0 15.4

Mean

oLOQ (%)

7.7 7.7 0 0 0 0

Median

Maximum

T1

T2a

T2b

T1

T2a

T2b

T1

T2a

T2b

NC NC NC NC 14.2 15.5

264.4 562.3 31.0 203.6 NC NC

274.1 571.2 38.7 209.7 NC NC

NC NC NC NC 13.3 15.4

0.0 0.0 0.0 0.0 NC NC

10.00 10.00 10.00 10.00 NC NC

NC NC NC NC 27.5 28.5

1730 3707 200.6 1423 NC NC

1730 3707 200.6 1423 NC NC

Abbreviations: LOD, limit of detection; LOQ, limit of quantification; NC, non concerned. T1: parameters are estimated after replacing values below LOD with LOD/2 and values lying between LOD and LOQ with (LOQLOD)/2; T2a: parameters are estimated after replacing o LOD values by zero and o LOQ values by LOD; T2b: parameters are estimated after replacing the censored data by the corresponding LOD or LOQ.

The contamination data are described in Table 2. As, for OAs and DAs, the censored data accounted for more 60% of values, zero and LOD (or LOQ) values were used in two separate estimations of the distributions and calculations (mean, median, percentilesy). On the other hand, as the censored values about SPXs were less than 60%, they were replaced by the half of the corresponding LOD or LOQ.

Exposure modelling General exposure model. An acute exposure corresponds to a short exposure to a harmful compound at high dose. Let us consider a phycotoxin denoted by m. The acute phycotoxin exposure is the amount of m ingested in a single meal. It is obtained by multiplying the edible portion size of one shellfish species by the concentration of m in this portion. For each phycotoxin, acute intakes were calculated individually for each shellfish species. Chronic exposure is a repeated exposure to low, or very low, doses for a long time. The chronic phycotoxin exposure is the amount of m ingested daily from the daily consumption of all shellfish species. The general exposure model used, here, to assess individual phycotoxin intake from shellfish consumption can be expressed as follows:18 Em ¼

P

ðCmj CRj Pj Þ BW

where Em is the individual exposure (mg/kg.bw/day) to the phycotoxin, m, from the ingested shellfish species, j, Cmj is the concentration (mg/kg flesh) of the same phycotoxin in the edible portion of the same species, CRj is the daily consumption rate (kg flesh/day) of this species, Pj is the proportion of a given shellfish species in a consumer diet (unitless), and BW is the consumer body weight (kg) assumed, in this study, to be 60 kg (as assumed by EFSA (EFSA, 2008; 2009; 2010)). The difference in acute- and chronic-exposure assessments stands in the consumption parameter to be used: the former takes into account the portion size of a given shellfish species, whereas the latter considers the daily consumption rate of all shellfish species. In this study, focus was on five of the bivalve species the most consumed in the geographical area under study: oysters, mussels, cockles, carpet shell clams and razor clams (king scallops were also considered in assessments about DA because such species are known to be the highest contaminated species by DA). Moreover, the approach in use for exposure calculation usually depends on the nature of the available data. This study was based on the two main methods, that is, deterministic and probabilistic approaches described in Kroes et al.3 and briefly recalled hereafter.

Deterministic approach (point estimate). In this approach, a fixed food consumption value, for example, the average or a high value, is multiplied by a fixed contamination level (often the average level or the upper & 2013 Nature America, Inc.

tolerance, or the permitted level depending on the legislation in force). Then, the intakes from all sources are summed. Point estimates are commonly used as a first step in exposure assessments because their calculations are rather simple and made at low cost.3,19,20 This strategy was developed, here, on assuming high phycotoxin concentrations in food as well as high shellfish consumption rates to determine whether additional and more in-depth exposure assessments were required.

Probabilistic approach. Given that a shellfish consumer will not eat, at each time, the same portion size and that the toxin level in the eaten portion will not be alike, the probabilistic calculation considers all of the combinations of occurrence and consumption data. Distributions for both the food consumption data and the contamination data were used in the model to simulate dietary intakes by repeatedly drawing random values for each input distribution. The description of input variables in terms of distributions allows one to characterise their variability and/or uncertainty. Monte Carlo simulation techniques are used by the model to generate output distributions of dietary intakes liable to be ultimately considered in probabilistic risk characterisation. Output distributions (i) give several exposure data (mean, median, minimum, maximum and all percentiles) and (ii) include a comprehensive analysis of the sensitivities of the resulting exposure with respect to uncertainties in parameters.3,19,20 Model assumptions The subjective assumptions in use in our simulation model were about factors liable to affect contamination rates such as regulatory limits, inter-species variability and cooking process. Because of their possible impact on the results obtained in exposure assessments, they have to be taken into account to generate the model outputs. Given that recreational shellfish harvesters can transgress bans, in this study, only the contamination distributions in purchased shellfish were right truncated at the regulatory limits when they exist (20 and 0.16 mg/g for DAs and OAs, respectively), whereas distributions of harvested shellfish contamination were not truncated (thus contamination higher than regulatory limits can be used). Each phycotoxin level is shellfish species specific, but the analyses were not made on all species. For each toxin, contamination levels were monthly determined in only two species. For the other ones, a 3-month preliminary study was conducted to gain insight into the variability of inter-species contamination. It allowed us to evaluate, for each toxin, the relationship between the levels of contamination in the most contaminated species and in the other ones, (for more details, see Picot et al.15). According to these results, a normal distribution was applied to the contamination levels in the species under study to describe the distributions of the levels in the non-analysed bivalve species.

Journal of Exposure Science and Environmental Epidemiology (2013), 435 -- 441

Probabilistic exposure assessment to phycotoxins Picot et al

438 Table 3. Acute and chronic dietary intakes of domoic acid, okadaic acid and spirolide obtained by a deterministic approach according to several scenarios for consumption and contamination. Contamination data

Acute assessment Consumption data (g/portion)

Type DA DA DA DA DA DA DA OA OA OA OA OA OA OA SPX SPX SPX SPX

Mean (T2a) Mean (T2a) Mean (T2b) Mean (T2b) Max (T2a, T2b) Regulatory limit Regulatory limit Mean (T2a) Mean (T2a) Mean (T2b) Mean (T2b) Max (T2a, T2b) Regulatory limit Regulatory limit Mean (T1) Mean (T1) Max (T1) Regulatory limit

Value (ng/g)

Type

g/ portion

562.3 562.3 571.2 571.2 3707 20,000 20,000 203.6 203.6 206.6 206.6 1423 160 160 26.2 26.2 116

Mean P95 Mean P95 P95 Mean P95 Mean P95 Mean P95 P95 Mean P95 Mean P95 P95

145.6 271.5 145.6 271.5 271.5 145.6 271.5 233.8 396.0 233.8 396.0 396.0 233.8 396.0 233.9 396.0 396.0

Chronic assessment ARfD (ng/kg.bw/ portion)

Exposurea (ng/kg.bw/ portion)

1365 2544 1386 2585 16,774 48,533 90,500 793 1344 805 1364 9392 623 1056 102.14 172.9 765.6

Consumption data

FAO/IOC/WHO

EFSA

Type

g/day

100,000

30,000

333.3

300

Mean P95 Mean P95 P95 Mean P95 Mean P95 Mean P95 P95 Mean P95 Mean P95 P95

5.93 17.9 5.93 17.9 17.9 5.93 17.9 10.4 22.8 10.4 22.8 22.8 10.4 22.8 5.93 17.9 17.9

No ARfD allocated

Exposurea (ng/kg.bw/ day)

55.6 167.3 56.5 169.9 1103 1977 5950 35.3 77.4 35.9 78.6 541 27.8 60.9 2.59 7.79 34.5

Abbrevition: ARfD, acute reference dose. For the deterministic approach, the phycotoxin intakes are calculated with consumption data concerning ‘‘eaters only’’. Contamination and consumption data concerned the bivalve species leading to the highest exposure: cockles for DA and mussels for OA and SPX. a Assuming a body weight equals to 60 kg.

As the phycotoxin levels are affected by the cooking process in use, this parameter has to be considered. The analyses were made on raw bivalves. Taking into account the cooking process impact, for each toxin, the ratio between the phycotoxin rates in raw samples and in cooked samples was determined in a preliminary study (for more details, see Picot et al.15). In this latter, it was shown that DA concentration was affected by the cooking process in very different ways in the five species studied. The cooking process led to a decrease in DA concentration in cockles (31%) and in razor clams (14%). On the contrary, DA concentrations were higher than uncooked samples in cooked mussels (46%), carpet shell clams (97%) and donax (17%). Concerning OAs and SPX, the lipophilic toxins, experiments were performed on raw and cooked mussels. For each toxin, an increase in toxin concentration is observed for cooked mussels (30%). In view of the lipophilic nature of these toxins, cooking effects on toxin concentrations should not be species-specific, as is the case for DA, thus we hypothesize that cooking impact is the same for all species. Then, a normal distribution describing the impact of cooking was assigned to the contamination levels of raw bivalves.

Model simulation The @Risk package, version 4.5 (Palisade, USA) with the Microsoft Excel spreadsheet under XP (Microsoft, USA) was used to perform risk analysis from Monte Carlo simulations and probability distributions so as to develop the exposure model on taking into account uncertainty and variability. Each simulation was run for 10,000 iterations to mimic the inherent uncertainty in shellfish contamination and consumption as well as the uncertainty in the mathematical process. The probability of existence of a phycotoxin in shellfish, its level in the shellfish and the probability of human exposure were all outputs of the mathematical model. To help in

the identification of critical points in the process, the model sensitivity was analysed. RESULTS AND DISCUSSION Acute and chronic exposures to each of the phototoxins under study conditions were assessed through application of both approaches. For the deterministic approach, the scenarios in use depended on the contamination and the consumption values used in the study: indeed, the contamination data were equal to either the mean, or the highest value obtained in our previous study (Picot et al.15) or the regulatory limit. The consumption data were obtained for the ‘eaters only’ and were equal to the mean or the 95th percentile.11 For the probabilistic approach, the consumption data were derived from the total population (including non-consumers). In this case, the exposure assessment model produced, for each phycotoxin, a probability density distribution of dietary intakes from all the bivalves under study. Acute-exposure assessment to phycotoxin(s) Acute exposure corresponds to the phycotoxin intake by an individual over a meal composed of a single portion of bivalves. For each bivalve species, the portion size is, thus, the quantity obtained by multiplying the consumption data by the contamination data. Deterministic exposure. The main results about exposure issued from the deterministic assessment are presented in Table 3. They deal only with the bivalve species leading to the highest exposure. Depending on the contamination and consumption values in use, the exposure levels are in the ranges 1.37 to 90.5 mg/kg.bw for DAs, 0.62 to 9.4 mg/kg.bw for OAs and 102 to 766 ng/kg.bw for

Journal of Exposure Science and Environmental Epidemiology (2013), 435 -- 441

& 2013 Nature America, Inc.

Probabilistic exposure assessment to phycotoxins Picot et al

439 Table 4.

Acute dietary intakes of domoic acid, okadaic acid and spirolide obtained by a probabilistic approach and comparison with toxicological reference values for each bivalve species.

Harvested oysters Harvested mussels Harvested cockles Harvested carpet chell clams Harvested razor clams Purchased oysters Purchased mussels Purchased cockles Purchased carpet chell clams Purchased razor clams ARfD

DA

OA

SPX

Exposurea (mg/kg.bw/portion)

Exposurea (ng/kg.bw/portion)

Exposurea (ng/kg.bw/portion)

Mean

Median

P95

Mean

Median

P95

Mean

0.16 0.52 0.70 0.74 0.14 0.15 0.61 0.02 0.03 0

0.06 0.20 0.27 0.29 0.05 0.06 0.24 0.01 0.01 0 100b/30c

0.65 2.09 2.75 3.02 0.53 0.61 2.41 0.07 0.13 0

19.4 456 702 378 133 17.7 183 4.00 3.80 0

7.62 151 243 149 44.9 7.1 79.0 1.7 1.5 0 333b/300c

78.1 1912 2808 1466 569.0 72.0 706 18.0 15.0 0

8.60 26.0 78.4 19.8 8.58 8.10 29.9 1.90 0.7 0

Median

P95

5.46 27.4 13.3 94.8 49.6 250 12.3 60.0 5.50 27.2 5.21 25.5 15.6 105 1.20 6.00 0.45 2.20 0 0 No ARfD allocated

Abbrevition: ARfD, acute reference dose. a assuming a body weight equals to 60 kg. b Proposed by FAO/IOC/WHO. c Proposed by EFSA.

SPX. These values are consistent with those found by the European Food Safety Authority (EFSA) panel (0.027 to 247.5 mg/ kg.bw for DAs, 0.103 to 4.67 mg/kg.bw for OAs and 60 ng/kg.bw for SPX).3--5 Deterministic estimates of dietary exposure are commonly used as a decision-help tool to decide whether further exposure assessments are required. The exposure levels found in this study prompted us to carry out more detailed and accurate investigations.

Concerning the SPX-exposure distribution, the highest mean, median and 95th percentile values were equal to 78.4, 49.6 and 250 ng/kg.bw, respectively, against 15, 7 and 60 ng/kg bw, for the values reported by the EFSA panel for acute exposure to SPX. These differences are mainly explained by the higher contamination rates in the region of interest compared with the distribution of contamination used by the EFSA panel built from levels reported by France, Italy and the Netherlands.

Probabilistic exposure. Table 4 presents the main results about exposure and comparison with toxicological reference values issued from the probabilistic assessment. For DAs, the exposure distribution obtained by multiplying a portion size distribution by a contamination distribution for each of the five shellfish species under study led to: (i) a mean up to 0.74 mg/kg.bw, (ii) a median value up to 0.29 mg/kg.bw and (iii) a 95th percentile value up to 3.02 mg/kg.bw. For mussels, the exposure values found further to the ingestion of purchased bivalves were greater than those from harvest, likely because of the higher portion size. In their semi-probabilistic assessment, the EFSA panel multiplied a triangular distribution of portion size (specified from the lowest-, the most probable and the highest-consumption values) by a distribution of contamination issued from a European data. This led to the finding of 2.17, 0.92 and 7.5 mg/kg.bw, for the mean, the median and the 95th percentile of acute intakes, respectively. It is worth noting the good agreement between the two sets of values. About OAs, the exposure distribution led to a mean value, a median value and a 95th percentile value up to 0.70, 0.24 and 2.81 mg/kg.bw, respectively. These results are of the same order of magnitude as those reported by the EFSA panel (0.23, 0.10 and 0.90 mg/kg.bw, for the mean, the median and the 95th percentile of acute intakes, respectively). On the other hand, our intake values proved to be one- to sevenfold for those proposed by the EFSA panel. These differences may come from the keeping, in our assessments, of the contamination rates above the regulatory limit for harvested bivalves, conversely to what was done in the study by ESFA panel. Moreover, as previous investigations had taught us that OA concentration in bivalves is elevated by the cooking process, its impact was taken into account here.

Acute-risk characterisation. For acute-risk characterisation, deterministic and probabilistic estimates of dietary exposure to phycotoxins have to be compared with the ARfD. The provisional ARfDs established by the JECFA (Joint FAO/WHO (Food and Agriculture Organization/World Health Organization) Expert Committee on Food Additives) are 100 and 0.33 mg/kg.bw for DAs and OAs, respectively, whereas no ARfD has been allocated for SPX (Toyofuku, 2006).21 In 2008 and 2009, the EFSA panel proposed ARfD values of 30 and 0.30 mg/kg.bw for DAs and OAs, respectively.4,5 In this study, all of the acute-exposure assessments about DA gave values less than the DA ARfD established by the JECFA and the EFSA. Concerning OAs, the highest probabilistic assessment (for harvested cockles) led to a mean exposure and a 95th percentile of, respectively, about 2.5-fold and 9-fold the OA ARfD, but to a median exposure almost 1.25-fold less than the OA ARfD. One should note that, for purchased bivalves, all exposures (means, medians and 95th percentiles) were below the OA ARfD, except for high consumers of mussels, they are above the ARfD. Characterising the SPX risk was impossible because of the unavailability of SPX ARfD value explained by the lack of quantitative data on acute oral toxicity by this phytotoxin.

& 2013 Nature America, Inc.

Chronic-exposure assessment to phycotoxins The chronic-exposure assessment corresponds to the level of exposure after a daily consumption of shellfish, thus the useful consumption data are the daily consumption rates. Deterministic exposure. Table 3 gives also the chronic-exposure levels issued from the deterministic exposure approach and shows that the deterministic exposure assessments led to maximal exposures of 5.95, 0.54 and 0.035 mg/kg.bw/day for DAs, OAs and SPX, respectively.

Journal of Exposure Science and Environmental Epidemiology (2013), 435 -- 441

Probabilistic exposure assessment to phycotoxins Picot et al

440 Table 5. Chronic dietary intakes of domoic acid, okadaic acid and spirolide obtained by a probabilistic approach and comparison with toxicological reference values for harvested, purchased and all bivalves. Censored values treatment

T2a or T1 (SPX) T2b

DA

OA

SPX

Exposurea (mg/kg.bw/day)

Exposurea (ng/kg.bw/day)

Exposurea (ng/kg.bw/day)

Bivalve sources

Harvested bivalves Purchased bivalves All bivalves Harvested bivalves Purchased bivalves All bivalves

Mean

Median

P95

Mean

Median

P95

Mean

Median

P95

0.07 0.04 0.11 0.07 0.04 0.11

0.05 0.03 0.09 0.05 0.03 0.09

0.18 0.13 0.26 0.20 0.14 0.28

44.4 9.70 54.1 45.9 10.3 56.2

29.3 5.10 39.0 30.9 5.40 41.1

134 34.5 149 137 36.6 155.0

3.5 1.9 5.4

2.8 1.3 4.6

8.3 5.7 11.9

Tolerable daily intake ARfD

No tolerable daily intake allocated 100b/30

c

333b/300c

No ARfD allocated

Abbreviation: ARfD, acute reference dose. a Assuming a body weight equals to 60 kg. b Proposed by FAO/IOC/WHO. c Proposed by EFSA.

Chronic risk characterisation. For chronic risk characterisation, deterministic and probabilistic estimates of dietary exposure to phycotoxins have to be compared with the tolerable daily intake (TDI). But, as no TDI has been allocated to phycotoxins by international committees, we compared our data about chronic dietary intakes of phycotoxins to the only available reference values, that is, ARfD. Table 5 shows that, according to the probabilistic approach, the mean and 95th percentile intakes for DA (0.11 and 0.26 mg/kg.bw/day, respectively) are about 100- and

Harvested oysters Harvested cockles Harvested razor clams Purchased mussels Purchased carpet shell clams

Harvested mussels Harvested carpet shell clams Purchased oysters Purchased cockles Purchased king scallops 34.1

36.6

40.0 35.0

0.0

8.2

7.2

2.5

0.8 0.4

11.3

9.5

15.6 1.8

0.2 0.2 0.0

1.6

3.8

5.1

7.7 0.4 1.0

2.6

5.0

6.6

10.0

11.3

15.0

18.5

20.0

20.2

18.7 22.0

25.0

25.9

25.9

30.0 % of contribution

Probabilistic exposure. Table 5 illustrates the comparison of the chronic-exposure levels issued from the probabilistic-exposure approach with the toxicological reference values for harvested-, purchased-bivales and ‘‘all bivalves’’. For DAs, it shows that, for ‘‘all bivalves’’, the mean value and the median are unchanged between T2a and T2b scenarios and equal to 0.11 and 0.09 mg/ kg.bw/day, respectively; the 95th percentile value is slightly different in T2a and T2b scenarios (0.26 and 0.28 mg/kg.bw/day, respectively). Concerning OAs, the ‘‘all bivalves’’-related exposure distribution presents maximal means of 54.1 and 56.2 ng/kg.bw/ day for T2a and T2b scenarios, respectively, as well as median values of 39.0 and 41.1 ng/kg.bw/day and 95th percentiles of 149 and 155 ng/kg.bw/day. One should note that the censored value scenario (T2a or T2b) has a very limited effect upon the chronic dietary exposure to phycotoxins. Concerning SPX, the chronic distribution of exposure (for ‘‘all bivalves’’) leads to a mean value of 5.4 ng/kg.bw/day, a median of 4.6 ng/kg.bw/day, and a 95th percentile of 11.9 ng/kg.bw/day. About the comparison of the contribution by harvested bivalves against the one by purchased bivalves, Table 5 shows clearly that, for DAs and SPX, the intakes derived from harvest are about 1.4- to 2.2-fold those derived from purchase. For OAs, the difference is even more marked (about fivefold) mainly because the contamination distribution of harvested bivalves took into account levels above the regulatory limit. Figure 1 illustrates the contribution of each bivalve shellfish to the daily intakes of each of the phycotoxin under study. It evidences that the three greatest contributor species are the cockles and the carpet shell clams from harvest as well as the mussels from both harvest and purchase. Their high contributions come from the high consumption and contamination rates for harvested species and to the high consumption rate for purchased mussels.

0.0 AD

AO

SPX

Figure 1. Contribution of each bivalve species to the daily intakes of domoid acid (DA), okadaic acid (OA) and spirolide (SPX) determined through the probabilistic approach (T2a and T1 treatments of censored values).

250-fold less than the most conservative ARfD. Despite the lack of TDI allocated to DA by international committees, Marie¨n22 estimated that a value of 75 mg/kg.bw provides a sound basis for a TDI. The daily DA intakes found in this study are, at least, about 300-fold less than the proposal by Marie¨n’s about TDI. For OA, the values of the mean and 95th percentile intake issued from the probabilistic approach (0.054 and 0.15 mg/kg.bw/day, respectively) are only about five- and twofold less than the most protective OA ARfD (0.30 mg/kg.bw). As neither ARfD nor TDI is available for SPX, no comparison was made. This study showed that, conversely to DA intakes, chronic OA intakes were close to ARfD. TDI is, by definition, less than ARfD. The former is, indeed, derived from a NOAEL (No Observed Adverse Effect Level) value or a LOAEL (Lowest Observed Adverse Effect Level) one determined from long-term toxicological studies, whereas the latter is determined from acute toxicological studies. Moreover, in addition to the traditional security factors employed for ARfD, the establishment of TDI requires the use of a few other ones such as, for example, an uncertainty factor of 10 to

Journal of Exposure Science and Environmental Epidemiology (2013), 435 -- 441

& 2013 Nature America, Inc.

Probabilistic exposure assessment to phycotoxins Picot et al

extrapolate subchronic to chronic exposure,23 leading to TDIs much lower than ARfD. The finding, in this study, of a chronic exposure to OA via shellfish consumption only two- to fivefold below the ARfD suggests that OA should be considered as a possible cause for concern about human health. Although no reference value has been allocated to SPX by international committees, the calculations made in this study highlighted the regular exposure of humans to low SPX doses. Thus, in the case where toxicological data indicate chronic impact by SPX on health, it would be worth taking into account the exposure to SPX.

CONCLUSION Further to the increasing number of reports about phytotoxininduced intoxications and deaths, these compounds have become a matter of concern for human health. But, phycotoxin-exposure assessments are almost non-existent because related data about consumption and contamination are missing. This led us to study, in the same geographical area, shellfish consumption by humans and shellfish contamination by phytotoxins to assess exposure of humans to these compounds. The acute- and chronic-exposure assessments made through both deterministic and probabilistic approaches showed that: (i) in terms of acute risk, monitoring programs and regulations about DA seem to be appropriate; on the other hand, OAs appear to be a cause for concern about high consumers in cases of high contamination levels that may exceed the OA ARfD. For instance, a high and ban-transgressing consumer could be exposed to an OA acute intake up to ninefold the ARfD; (ii) about chronic risk, the finding of daily OA intakes close to the ARfD, known to be, by definition, much greater than the TDI, suggests that, among the phycotoxins under study, OA is the one to be considered. Moreover, it should be noted that bivalves contain regularly SPX at low concentrations. Chronic and subchronic data on SPX are missing, but in case of (sub)chronic toxicity, SPX exposure should be taken into consideration. These phycotoxin-exposure assessments were aimed at making a first realistic evaluation of human exposure to phycotoxins. Their interest stands in the facts that: (i) they were based on consumption and contamination data in the same subpopulation and area, (ii) the recreational shellfish harvesters under study constitute an at-risk subpopulation, (iii) inter-species variability in contamination and consumption data was taken into account, (iv) the impact of cooking process on phycotoxin levels was also considered. To gain more comprehensive insight into this health issue, in the future, it would be worth: (i) increasing the number of shellfish species to be investigated, (ii) considering the contamination data relative to recorded cases of intoxication further to ingestion of fish and crustaceans, (iii) extending the contamination database to several years and (iv) studying co-exposure to several phycotoxins.

CONFLICT OF INTEREST The authors declare no conflict of interest.

ACKNOWLEDGEMENTS We thank Dr M.P. Friocourt for assistance in the English writing of this paper.

& 2013 Nature America, Inc.

REFERENCES 1 Shumway S.E., Van Egmond H.P., Hurst J.W., and Bean L.L. Management of shellfish resources. In: Hallegraeff GM, Anderson DM, Cembella AD. (Eds). Manual on Harmful Marine Microalgae. IOC Manuals and Guides No. 33. Intergovernmental Oceanographic Commission of UNESCO: Paris France, 1995, 433--474. 2 Van Dolah F.M. Marine algal toxins: origins, health effects, and their increased occurrence. Environ Health Perspect 2000: 108: S133--S141. 3 Kroes R., Mu¨ller D., Lambe J., Lo¨wik M.R.H., Van Klaveren J., and Kleiner J., et al. Assessment of intake from the diet. Food Chem Toxicol 2002: 40: 327--385. 4 Opinion of the Scientific Panel on Contaminants in the Food chain on a request from the European Commission on marine biotoxins in shellfish---okadaic acid and analogues. EFSA J 2008: 589: 1--62. 5 Scientific Opinion of the Panel on Contaminants in the Food Chain on a request from the European Commission on marine biotoxins in shellfish---domoic acid. EFSA J 2009: 1181: 1--61. 6 Scientific Opinion on marine biotoxins in shellfish---Cyclic imines (spirolides, gymnodimines, pinnatoxins and pteriatoxins). EFSA J 2010: 1628: 1--39. 7 Picot C., Nguyen T.A., Roudot A.C., and Parent-Massin D. A preliminary risk assessment of human exposure to phycotoxins in shellfish: a review. Hum Ecol Risk Assess 2011: 17: 328--366. 8 Burger J., Sanchez J., and Gochfeld M. Fishing, consumption, and risk perception in fisherfolk along an east coast estuary. Environ Res 1998: 77: 25--35. 9 Gagnon F., Tremblay T., Rouette J., Cartier J. Chemical risks associated with consumption of shellfish harvested on the north shore of the St Lawrence River’s lower estuary. Environ Health Perspect 2004: 112: 883--888. 10 Leblanc J.C. CALIPSO, Fish and seafood consumption study and biomakers of exposure to trace elements, pollutants and Omega-3. AFSSA (French food safety agency) 2006 (available at: http://www.afssa.fr/Documents/PASER-Ra-Calipso.pdf). 11 Picot C., Nguyen T.A., Carpentier F.G., Roudot A.C., and Parent-Massin D. Relevant shellfish consumption data for dietary exposure assessment among high shellfish consumers, Western Brittany, France. Int J Environ Health Res 2011: 21: 86--105. 12 Picot C. Evaluation du risque lie´ a` l’exposition aux phycotoxines via la consommation de coquillages chez les peˆcheurs a` pied du Finiste`re. PhD manuscript, Universite´ de Bretagne Occidentale, Brest, France 2010. 13 Guidelines for the Study of Dietary Intakes of Chemical Contaminants, WHO Offset Publication: Geneva, 1985, 87: 104p. 14 Todd E.C. Chronology of the toxic mussels outbreak. Can Dis Wkly Rep 1990: 16: S3--S4. 15 Picot C., Limon G., Durand G., Wesolek N., Parent-Massin D., and Roudot A.C. Domoic acid, okadaic acid and spirolides: inter-species variability in contamination and cooking effects. Food Public Health 2012: 2: 50--57. 16 Gerssen A., McElhinney M.A., Mulder P.P.H., Bire R., Hess P., De Boer J. Solid phase extraction for removal of matrix effects in lipophilic marine toxin analysis by liquid chromatography-tandem mass spectrometry. Anal Bioanal Chem 2009: 394: 1213--1226. 17 GEMS/Food-EURO Second Workshop on Reliable Evaluation of Low-Level Contamination of Food, 1995, Kulmbach: Germany, EUR/ICP/EHAZ.94.12/WS04-FSR/ KULREP95. 18 Guidance for assessing chemical contaminant data for use in fish advisories. Risk Assessment and Fish Consumption Limits 2000: Vol 2, 3rd Edn. EPA 823-B-94008.US Environmental Protection Agency, Office of Water, Washington, DC. 19 Counil E., Verger P., and Volatier J.L. Handling of contamination variability in exposure assessment: a case study with ochratoxin A. Food Chem Toxicol 2005: 43: 1541--1555. 20 Tressou J., Cre´pet A., Bertail P., Feinberg M.H., and Leblanc J.C. Probabilistic exposure assessment to food chemicals based on extreme value theory. Application to heavy metals from fish and sea products. Food Chem Toxicol 2004: 42: 1349--1358. 21 Toyofuku H. Joint FAO/WHO/IOC activities to provide scientific advice on marine biotoxins (research report). Mar Pollut Bull 2006: 52: 1735--1745. 22 Marie¨n K. Establishing tolerable dungeness crab (Cancer magister) and razor clam (Siliqua patula) domoic acid contaminant levels. Environ Health Perspect 1996: 104: 1230--1236. 23 Lewis S.C. Subchronic to chronic exposure extrapolation: Toxicologic evidence for a reduced uncertainty factor. Hum Ecol Risk Assess 1995: 1: 516--526.

Journal of Exposure Science and Environmental Epidemiology (2013), 435 -- 441

441