Environment International 66 (2014) 107–114 Contents lists available at ScienceDirect Environment International j ourna l homepage: www.e lsev ie r .com/ locate /env int Cooking fish is not effective in reducing exposure to perfluoroalkyl and polyfluoroalkyl substances Satyendra P. Bhavsar a,b,⁎, Xianming Zhang b, Rui Guo a, Eric Braekevelt c, Steve Petro a, Nilima Gandhi b, Eric J. Reiner a,b, Holly Lee a, Roni Bronson c, Sheryl A. Tittlemier c,1 a Ontario Ministry of the Environment, Toronto, ON M9P 3V6, Canada b University of Toronto, Toronto, ON M5S 3E8, Canada c Health Canada, Ottawa, ON K1A 0L2, Canada ⁎ Corresponding author at: Ontario Ministry of the Monitoring and Reporting Branch, 125 Resources Road, Tel.: +1 416 327 5863; fax: +1 416 327 6519. E-mail addresses: satyendra.bhavsar@ontario.ca, s.bhav 1 Current affiliation:GrainResearch Laboratory, Canadia MB R3C 3G8, Canada. 0160-4120/$ – see front matter. Crown Copyright © 2014 http://dx.doi.org/10.1016/j.envint.2014.01.024 a b s t r a c t a r t i c l e i n f o Article history: Received 17 November 2012 Accepted 23 January 2014 Available online 19 February 2014 Keywords: Perfluoroalkyl substances Fish consumption Cooking PFOS Emerging contaminant Great Lakes Consumption of fish is considered a part of a healthy diet; however, health risks from fish consumption exist due to potential exposure to various contaminants accumulated in fish. Cooking fish can reduce exposure to many organic chemicals in fish. Similar results have been presented for low levels of perfluoroalkyl and polyfluoroalkyl substances (PFASs), a class of contaminants of emerging concern, in grocery store fish. We examined the effec- tiveness of three cookingmethods (i.e., baking, broiling, and frying) on reducing PFAS levels in four sport fish spe- cies. Samples of Chinook salmon, common carp, lake trout andwalleyewere collected from four rivers in Ontario, Canada and skin-off fillets were analyzed for regular groups of PFASs such as perfluoroalkyl carboxylic acids (PFCAs) and perfluoroalkane sulfonic acids (PFSAs), as well as perfluoroalkyl phosphonic acids (PFPAs), perfluoroalkyl phosphinic acids (PFPIAs) and polyfluoroalkyl phosphoric acid diesters (diPAPs), which are PFASs of emerging concern. Perfluorooctane sulfonate (PFOS) was the dominant PFAS detected and the concen- trationsweremore than an order ofmagnitude higher than those reported for fish fromgrocery stores in Canada, Spain, and China. Although concentrations of PFOS infishfillets generally increase after cooking, amounts of PFOS largely remain unchanged. Relatively minor differences in changes in the fish PFAS amounts after cooking depended on fish species and cooking method used. We conclude that cooking sport fish is generally not an effective approach to reduce dietary exposure to PFASs, especially PFOS. Crown Copyright © 2014 Published by Elsevier Ltd. All rights reserved. 1. Introduction Fish consumption has been regarded as beneficial to human health due to the high-quality protein, minerals, antioxidants, vitamins and omega-3 polyunsaturated fatty acids (PUFAs) in fish (Domingo, 2007). Health benefits of fish consumption include optimal brain and retinal development and reduced risk of coronary heart disease (Egeland and Middaugh, 1997). Several health agencies such as the World Health Organization (WHO, 2002) and Health Canada (Health Canada, 2007a) recommend at least two servings of fish per week. While enjoying health benefits of fish consumption, humans are also subject to poten- tial health risks from exposure to various contaminants accumulated in fish (Mozaffarian and Rimm, 2006). In order to minimize such health risks, various government agencies determine safe amounts of fish that can be consumed, and as necessary, issue fish consumption advisories Environment, Environmental Toronto, ON M9P 3V6, Canada. sar@utoronto.ca (S.P. Bhavsar). nGrain Commission,Winnipeg, Published by Elsevier Ltd. All rights (Bhavsar et al., 2011; Health Canada, 2007b; OMOE, 2011; U.S. EPA, 2010). It has been demonstrated that cooking and removal of skin can re- duce the concentrations of some organic contaminants, such as dioxins and polychlorinated biphenyls (PCBs) (Hori et al., 2005; Sherer and Price, 1993; Zhang et al., 2013). However, such practices have minimal impact on fish concentrations of some other contaminants, such as heavy metals. In fact, some studies have reported increases in concen- trations of heavy metals after skin removal or cooking (Morgan et al., 1997; Perelló et al., 2008). The extent to which the concentration of a contaminant may be altered by cooking processes depends upon the type of tissue in which it accumulates. Neutral organic contaminants generally have a higher affinity for the fatty tissues of fish (Bertelsen et al., 1998), and loss of fat via skin removal or cooking is a major con- tributor to reducing concentrations of these contaminants (Bayen et al., 2005; Wilson et al., 1998). In contrast, heavy metals generally bind with tissue proteins (Hamza-Chaffai et al., 1995) and are less affected by fish processing methods. Similar to heavy metals, binding to proteins is also considered to be the bioaccumulation mechanism for the ionizable perfluorinated com- pounds, a group of organic compounds with unique surface properties and low water and oil solubility (Conder et al., 2008; Haukas et al., reserved. http://crossmark.crossref.org/dialog/?doi=10.1016/j.envint.2014.01.024&domain=pdf http://dx.doi.org/10.1016/j.envint.2014.01.024 mailto:satyendra.bhavsar@ontario.ca mailto:s.bhavsar@utoronto.ca http://dx.doi.org/10.1016/j.envint.2014.01.024 http://www.sciencedirect.com/science/journal/01604120 108 S.P. Bhavsar et al. / Environment International 66 (2014) 107–114 2007; Jones et al., 2003; Kelly et al., 2009). Historically, they have been used in diverse applications including surfactants and in aqueous film forming foam fire fighting agents (AFFF). The strong carbon-fluorine bonds of PFASs make them less susceptible to degradation and highly persistent in the environment (Lindstrom et al., 2011). The persistence makes it possible for PFASs to undergo long range transport to remote regions where no PFASs have been produced or used (Houde et al., 2011; Wania, 2007). Further, PFASs have been found in organisms at various trophic levels (Houde et al., 2011), and are considered endo- crine disruptive, immunotoxic and tumorgenic in laboratory animals at relatively high doses (Betts, 2007). Because of their persistent nature, long range transport capability, bioaccumulation potential and toxicity, a group of PFASs including perfluorooctanesulfonic acid (also known as perfluorooctane sulfonate or PFOS), its salts and perfluorooctane sulfo- nyl fluoride were listed in Annex B of the United Nations Stockholm Convention on Persistent Organic Pollutants list in 2009 (UNEP, 2009). In the past decade, increasing surveillance data of PFAS concentra- tions in fish have become available worldwide. In some parts of North America, surveillance of sport fish for PFASs has resulted in the issuance of restrictive fish consumption advisories (Delinsky et al., 2010; OMOE, 2011). However, in contrast to neutral organic contaminants such as PCBs and dioxins, little information on the effects of different cooking methods on PFAS levels in fish is available. A recent comparison of PFAS concentrations in raw fish with the corresponding cooked fish by Del Gobbo et al. (2008) showed a decline in PFAS concentrations after cooking. These findings are, however, in contrast to the unchanged or even increased concentrations of heavy metals in fish after cooking even though both PFASs and heavy metals bind to protein. Since the fish samples utilized byDel Gobbo et al. (2008)were obtained fromgro- cery stores and the PFAS concentrations were relatively low (e.g., PFOS ranging 0.21–1.68 ng/g ww) compared to those for sport fish, it is pos- sible that analytical uncertaintymight have contributed to the proposed decline in the concentrations after the cooking. Therefore, more data on how different cooking methods affect PFAS concentrations in fish are needed to refine exposure assessments and ensuing fish consumption advice. The goal of this study is to gather information on the effectiveness of three cooking methods (i.e., baking, broiling, and frying) to reduce PFC levels in four fish species sampled from rivers in Ontario, Canada, rather than fromagrocery store.We consideredChinook salmon (Oncorhynchus tshawytscha), common carp (Cyprinus carpio), lake trout (Salvelinus namaycush) and walleye (Sander vitreus). In addition to perfluoroalkyl carboxylic acids (PFCAs) and perfluoroalkane sulfonic acids (PFSAs), we also examined perfluoroalkyl phosphonic acids (PFPAs), perfluor- oalkyl phosphinic acids (PFPIAs) and polyfluoroalkyl phosphoric acid diesters (diPAPs), which are PFASs of emerging concern. PFPAs and PFPIAs have been used as wetting and foam-dampening agents, while diPAPs have been used in food-contact paper products (Begley et al., 2005; Dupont, 2012; Mason Chemical Co., 2012). However, to date no submission has been made to the Food Directorate of Health Canada requesting the use of diPAPs in food packaging materials sold in Canada (Rulibikiye, Health Canada, personal communication). 2. Methods 2.1. Sample collection and preparation Fish sampleswere collected in summer/fall of 2010/11 from four riv- ers inOntario, Canada: Credit River (Chinook salmon,N=5, 58–91 cm), Thames River (common carp, N = 5, 66–76 cm), Niagara River (lake trout, N = 4, 58–79 cm), and Welland River (walleye, N = 5, 46–62 cm). Relatively elevated concentrations of PFASs in fish from these locations were expected based on nearby industrial activities or previous monitoring work conducted by the Ontario Ministry of Envi- ronment (OMOE). The samples were measured for total length and weight, sexed and filleted (both sides, skin-off) in the field. The fillets were stored on ice and transported to the Ontario Ministry of the Environment's Sport Fish Contaminant Monitoring Program office in Toronto, Ontario, Canada where they were stored at −20 °C until further processing. The two fillets from each fish were partially thawed and segmented into 16 parts and classified into four groups (raw plus three cooking methods) as shown in the Supporting Information (SI) Fig. S1 in order to minimize influence of potentially varying PFAS levels in different parts of the fillets on the study results. Four subsamples from each fish were stored at −20 °C until further processing. Fillet samples for three cooking methods were then stored on ice and transported to the Health Canada laboratory in Ottawa, Ontario, Canada where they were stored at −20 °C until cooked. Raw and cooked fillets were later homogenized at the Toronto laboratory using a Buchi B-400Mixer and stored again at−20 °C until chemical analysis. 2.2. Cooking details Frozen fish fillets were allowed to warm to room temperature. The weights of large aluminum weighing dishes were measured, then 10 g canola oil was added to each dish and evenly distributed over the bot- tom of the dish using a silicone brush. We chose canola oil because it is typically used for cooking in Canadawhere the experimentswere per- formed, and is the third most consumed vegetable oil in the world (Canola Council of Canada, 2013). The weight of the dish with the oil was measured and recorded. Each fish sample was then placed in its labeled weighing dish. The total weight (dish + oil + fish) was also measured. 2.2.1. Frying An electric frying pan was set to 175 °C and given 10 min to reach test temperature. The aluminum dishes were placed in the frying pan and cooked uncovered. After 5min, the fish fillets were carefully flipped with a plastic spatula and cooked for an additional 5 min. 2.2.2. Baking A small toaster oven was preheated to 200 °C (measured using an oven thermometer). The aluminum dishes were placed in the oven and cooked uncovered for 15 min. 2.2.3. Broiling The toaster oven was set to broil. The broiling temperature (mea- sured using an oven thermometer) was set at 300 °C. The aluminum dishes were placed in the oven and cooked uncovered for 10 min. 2.2.4. Post-cooking The samples were removed from heat and the internal temperature of the fish was immediately measured with a digital probe. The fish were allowed to cool before the total weight (dish + oil + fish) was measured. The fish was removed from its weighing dish, wrapped in aluminum foil, replaced in its original labeled bag and frozen to −20 °C for later analysis. The final weight of the dish with cooking juices and leftover canola oil was also measured. The weights of the cooking juices generated were calculated by subtracting pre-cooking weight of dish with oil from the final weight of the dish with juices and oil. Cooking juices and leftover canola oil were transferred to a polypropyl- ene sample bottle and frozen for later analysis. 2.2.5. Blanks Canola oil (10 g) was added to an unused aluminum weighing dish and evenly distributed over the bottom of the dish using a silicone brush. The dish was then baked at 200 °C for 15 min. The dish was allowed to cool, then the oil was transferred to a polypropylene sample bottle. 109S.P. Bhavsar et al. / Environment International 66 (2014) 107–114 2.3. Sample preparation for PFAS analysis Complete details of sample preparation and PFAS analysis including the compounds analyzed are provided by Guo et al. (2012). Briefly, 3mL of acetonitrilewas added to a 1 g homogenized fillet sample and 1mL of 60/40 acetonitrile/water was added to cooking juices or blank (canola oil). The resulting solution was shaken, centrifuged and transferred to a 15 mL polypropylene tube. The sample was extracted with a second aliquot of solvent and the supernatant was combined with the original solvent. Acetonitrile extracts were evaporated to about 1 mL using a gentle stream of nitrogen. The sample was then extracted twice with 5 mL methyl t-butyl ether (MTBE) following addition of 1 mL of 0.5 M tetrabutyl ammonium hydrogen sulphate (TBAS) at pH 4. TheMTBE al- iquots were evaporated to dryness under a gentle stream of nitrogen and reconstituted in 1 mL methanol. Prior to instrumental analysis, the extract was split. One fraction was spiked with mass-labeled PFCAs and PFSAs, and diluted with HPLC-grade water to make the sam- ple extract solvent mixture 60% methanol and 40% water (V/V). The other fraction was spiked with mass-labelled diPAPs and was not dilut- ed with water. Both fractions were filtered with a 0.2 μm syringeless polypropylene filter. Fig. 1. Percent of fishmass change after baking, broiling and frying of the Chinook salmon, common carp, lake trout and walleye fillets. The results are presented as box–whisker plots, where the boxes represent the 25th–75th percentiles; the lines in the boxes repre- sent the medians; the whiskers represent non-outlier ranges; and the hollow dots repre- sent outliers and the stars represent extreme values. 2.4. Instrumental analysis Analysis was performed using an Agilent 1200LC liquid chromato- graph with a Restek Ultra C18 column (50 × 2.1 mm, 3 μm; Restek, Bellefonte, PA, USA) coupled to a 4000 QTrap triple-quadrupole mass spectrometer (Applied Biosystems/MDS Sciex) in negative ion electrospraymode usingmultiple reactionmonitoring (MRM). Themo- bile phase flow rate was 250 μL/min. Gradient elution was used to sep- arate each group of analytes. For the PFCAs and PFSAs, the gradient started with 65% methanol with 10 mM ammonium acetate (Solvent A) and 35%HPLC-gradewater with 10mMammoniumacetate (Solvent B), followed by a linear gradient to 90% A in 6 min, kept for 1 min and then returned to the initial composition. The column temperature was 35 °C. For the PFPAs and PFPIAs, the gradient started with 20% A follow- ed by a linear gradient to 65%A in 1min, followed by a linear gradient of solvent A to 95% in 4 min. The final composition (95% A) was held for 2min before being returned to initial composition. The column temper- ature was 30 °C. For the diPAPs, the gradient beganwith 70% A, the gra- dient was then changed to 95% A in 0.1 min, followed by linear gradient of solvent A to 100% in 1 min. This composition was kept for 4 min and then returned to the initial composition. The column temperature was held at 25 °C. Multiple reaction monitoring transitions and optimized mass spectrometer parameters for each target compounds are summa- rized by Guo et al. (2012). PFCAs, PFSAs and diPAPs were quantified by multi-point (5–1000 fg/μL) calibration using mass-labeled internal standards, and transitions and conditions described by Guo et al. (2012). PFPIAs were quantified using external standard calibration as mass labeled internal standards were not available and matrix suppression/enhancement was not observed (Guo et al., 2012). PFPAs were quantified by standard addition as no internal standard was available and matrix effects were observed (Guo et al., 2012). Two procedural blanks (HPLC-grade water) and a spiked fish blank (Captain High Liner, Alaskan Pollock) were analyzed with each set of 15 to 20 fish samples as quality control checks. Recoveries in spiked samples typically range between 80 and 120% (Guo et al., 2012). Values below limits of quantification (LOQ) were reported as not detected (ND). Each sample was extracted once and analyzed twice, and values were reported as arithmetic means of the two injec- tions. Perfluorooctanoic acid (PFOA) and 6:2 diPAP were detected in procedural blanks with a mean concentration of 12 pg/g for PFOA and at lower than the LOQ for 6:2 diPAP. Values for PFOA were blank- corrected. There are no certified referencematerials available for cooked fish to ideally test if recoveries from cooked fish are the same as from raw. However, recoveries of internal standard added to raw and cooked fish and seafood in Del Gobbo et al. (2008) did not differ (unpublished data), indicating that the state of the fish (i.e., raw vs. cooked) does not affect fortified analyte recovery. In addition, Tittlemier et al. (2007) show that recoveries of PFASs for a variety of fortified food ma- trices including cooked chicken nuggets, organ meats, cured pork, and infant cerealwere similar suggesting that there are nodifferences across these matrices. 2.5. Data analysis The changes in fish mass and PFOS concentrations/amounts were examined using two-factor (fish species and cooking method) ANOVA withmultiple comparisons (TukeyHSDmethod). Statistical significance was set at p b 0.05. Self organizing map (SOM) was applied to the log- transformed concentration profiles of ten PFASs in five different fish species subjected to four processing methods (raw plus three types of cooking). SOM is a type of unsupervised artificial neural network which generates simple geometric relationships on a low (two-) dimen- sional display from a high-dimensional input space of the training sam- ples (Kohonen, 2001). SOM has been widely used as a clustering and projection technique to extract features from large high-dimensional datasets for visualization and interpretation (Ericson et al., 2007; Mari et al., 2010). In order to generate the SOM, a value for each dimension of each member of samples is needed. Therefore, if a PFAS was not de- tected in all samples of a group (fish species, processing method), the PFAS was not included in construction of the SOM. In the case when a PFASwas detected only in a part of the sampleswithin a group (fish spe- cies, processing method), half of the detection limit was assigned to the non-detected PFAS. After processes of training and self-learning from the input datasets, SOMswith an array of nodes (neurons) were gener- ated using the SOM toolbox (version 2) (Alhoniemi et al., 2005). A weight vector of the same dimension of the input data (10 PFASs) was associated with each neuron. SOM can be separated into component planes which represent each dimension of the weight vector. Compo- nent planes can thus display the variable contribution to each neuron in the map. 110 S.P. Bhavsar et al. / Environment International 66 (2014) 107–114 3. Results and discussion 3.1. Changes in sample mass after cooking The percentages of changes inmass for each fish species and cooking method are displayed using box–whisker plots (Fig. 1). Overall, cooking decreased the mass of the fish fillets by 10–25%. The mass declines dif- fered significantly (p b 0.01) among species/cooking combinations; however, variations for the replicated samples of a species/cooking combination ranged only 3–10% (Fig. 1). There were significant differ- ences (p b 0.05) in average declines of fish mass between lake trout (12.5%) and the other three species (17–19%) (Fig. S2, Table S1). For the three cooking methods, average fish mass decline due to broiling (21%) was statistically higher (p b 0.01) than those due to baking (16%) and frying (15%) (Table S2). The two-factor ANOVAalso indicated that magnitude of fish mass change depends on both species and cooking method (Fig. S2). The change infishmassduring baking, broiling and frying canbemain- ly attributed to the loss ofmoisture, which contributes approximately 70% Fig. 2. Concentrations (ng/gwetweight) of perfluorinated compounds in raw, baked, broiled an standard deviation of replicated samples. to wet weight of a fish fillet. Previous studies (Gokoglu et al., 2004; Weber et al., 2008) observed less moisture loss by baking than grilling (equivalent to broiling of our study), which is consistent with the find- ings of this study (Fig. 1). However, reported higher dehydration of fish fillets due to frying compared to baking and grilling (Weber et al., 2008) is not supported by the lower mass reduction of fried fish samples in this study. This discrepancy is possibly due to fish fillets absorbing some oil while losing some moisture during frying. 3.2. PFAS concentrations and profiles Five groups of PFASs, namely PFCAs, PFSAs, PFPAs, diPAPs and PFPIAs, were analyzed in raw, baked, broiled and fried fish fillets. Of the PFCAs analyzed, perfluoroheptanoic acid (PFHpA) was not detected in any sample and PFOAwas detected in only 50% of the samples. PFPAs were not detected in any fish samples, and diPAPs were detected in b50% of the samples. Further, 8:8 PFPIA was detected in only 25% of the samples. These results are consistent with the recent findings of Guo et al. (2012) for Great Lakes lake trout. Therefore, in the cooking d fried Chinook salmon, common carp, lake trout andwalleyefillets. Error bars indicate the image of Fig.�2 111S.P. Bhavsar et al. / Environment International 66 (2014) 107–114 effect analysis, PFHpA, PFOA, PFPAs, diPAPs and 8:8 PFPIAwere excluded. The PFASs that were detected in N60% of the samples included per- fluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluor- oundecanoic acid (PFUnDA), perfluorododecanoic acid (PFDoDA), perfluorotridecanoic acid (PFTrDA), perfluorotetradecanoic acid (PFTeDA), perfluorohexane sulfonic acid (PFHxS), PFOS, perfluorodecane sulfonic acid (PFDS), 6:6 PFPIA and 6:8 PFPIA (Table S3). Of the PFASs analyzed, PFOS concentrations (16–53 ng/g ww) were 1–2 orders of magnitudes higher than those of the other PFASs (Fig. 2, Table S3), which is consistent with the PFAS profiles in fish measured by other studies (Awad et al., 2011; Del Gobbo et al., 2008; Zhang et al., 2011). PFOS concentrations in fish analyzed in this study were compared with other studies which measured PFOS in fish muscle or edible portion. Fish PFOS levels measured in this study were within the ranges of b6–300 ng/g ww for fish from Michigan waters (Giesy and Kannan, 2001), 1–2000 ng/g ww for fish from Minnesota lakes and rivers (Delinsky et al., 2010), and b1–100 ng/g ww for fish from the Great Lakes (De Silva et al., 2010); however, they were more than an order of magnitude higher than those for fish collected from retail markets in Canada (Del Gobbo et al., 2008; Tittlemier et al., 2007), Spain (Ericson et al., 2008), and China (Gulkowska et al., 2006). As such, this study addresses the current knowledge gap on the effects of cooking on environmentally relevant fish PFOS levels, especially for the Great Lakes area. Concentrations of∑PFCAs varied between fish species and cooking methods, and contributed only 2–20% of∑PFASs (Table S3). Unlike the measurements of PFNA and PFUnDA as the two dominant PFCAs by Del Gobbo et al. (2008), PFNA in this studywas lower than the longer chain PFCAs for species other than lake trout. PFUnDA was at higher concen- trations than PFNA, but depending on fish species, PFUnDA could be lower than PFDoDA and PFTeDA. Previous studies indicated PFCAs with an odd number of perfluoroalkyl carbons had higher concentra- tions than the corresponding shorter, even number PFCAs in fish (Del PFNA PFDoD 3.47 4.06 4.65 PFOS PFDS Labels a1 a1 c1 c4 c1 c1 a4 c1 a1 c2 a1 d1 d1 d4 d4 d1 d2 b1 b2 b1 b1 0.52 1.54 2.56 PFHxS 0.2 2 3.8 U-matrix Fig. 3. U-matrix, unit labels and component planes of self-organizing maps for the logarithm tr (a) Chinook salmon, (b) common carp, (c) lake trout and (d) walleye. Gobbo et al., 2008; Martin et al., 2004). In this study, we found that the pattern of relative abundance of PFASs with odd–even number of carbons varies among the fish species, and PFCAs in only lake trout and walleye followed such a pattern. PFPIAs differed amongfish specieswith the average concentration in common carp being about an order of magnitude higher than the other fish species (Table S3). 6:6 PFPIA and 6:8 PFPIA were detected in 70– 75% of all the fish samples, while 8:8 PFPIA was detected only in com- mon carp. 6:8 PFPIA was dominant in all five fish species with concen- trations ranging from 0.01 to 0.3 ng/g ww in raw samples. As a new class of PFASs, PFPIAs have been reported in few fish studies. The con- centrations of 6:6 PFPIA and 6:8 PFPIA observed for raw fish in this study (Table S3) were either comparable or higher than those for Great Lakes lake trout recently reported by Guo et al. (2012). Our find- ing of higher 6:8 PFPIA concentrations than 6:6 PFPIA is also consistent with those for Great Lakes lake trout (Guo et al., 2012) and human sera (Lee andMabury, 2011), and could reflect differences in emissions and/ or bioaccumulation potential between these two PFPIAs. 3.3. PFAS patterns and clustering In order to investigate how different cooking methods affect PFAS concentrations in the four fish species, the four dimensional datasets (cooking methods, fish species, chemicals, concentrations) were ana- lyzed using the self-organizingmap (SOM), anunsupervisedneural net- work technique to visualize high dimensional data and to establish pattern similarities (Kohonen, 2001). The color of a neuron in the U-matrix of SOM indicates the average distance between the neuron and its closest neighbors. Therefore, cluster of the dataset can be rep- resented by the color groups in the U-Matrix. The U-matrix and Labels of the SOM for the four dimensional datasets clustered the values by fish species and not by cookingmethods (Fig. 3), indicating that baking, broiling and frying are not unique from each other in changing PFAS 1.26 1.92 2.58 1.39 2.13 2.87 PFDA 1.21 2.06 2.91 PFUnDA 1.17 2.12 3.07 A 1.41 2.19 2.97 PFTrDA 0.86 1.99 3.12 PFTeDA 1.76 2.44 3.12 0.02 0.99 1.96 6:6 PFPIA 0.04 1.17 2.30 6:8 PFPIA ansformed PFAS concentrations in (1) baked, (2) broiled, (3) fried, and (4) raw samples of image of Fig.�3 112 S.P. Bhavsar et al. / Environment International 66 (2014) 107–114 concentrations in fish. Component planes (c-planes) represent the con- tribution of each variable to each neurons of the SOM. The c-planes can identify the PFASs with similar contribution to the data clustering. As such, similar color distributions in the c-planes of PFDA and PFUnDA as well as 6:6 PFPIA and 6:8 PFPIA indicate positive correlations be- tween the PFASs (Fig. 3). The c-planes and unit labels suggest that PFPIAs contributed mainly to the clustered neurons representing com- mon carp (bottom right of the SOM) while PFOS contributed mainly to the clustered neurons representing walleye (near top right of the SOM). Such information indicates that PFPIAs were higher in common carp and PFOS was higher in walleye compared to the other species. 3.4. Effect of cooking on PFASs in fish Because PFOS is generally the most abundant PFAS reported in fish (e.g., Fig. 2) and the presence of which has required restrictive fish con- sumption advisories in at least some parts of North America (Delinsky et al., 2010; OMOE, 2011), below we present detailed analysis of the changes in the fish PFOS concentrations and amounts after baking, broiling and frying. PFOS concentrations in all fish species increased significantly (p b 0.05) after cooking except for broiling and frying of common carp, which had no significant changes (p N 0.05) in PFOS concentra- tions (Fig. 4a). This is in contrast to reported 54–100% decline in low PFAS concentrations after cooking fish obtained from a grocery store Fig. 4. Relative changes (in percentage) of PFOS (a) concentrations and (b) amounts after baking, broiling and frying of the Chinook salmon, common carp, lake trout and walleye fillets compared to the corresponding raw fish values. The results are presented as box– whisker plots, where the boxes represent the 25th–75th percentiles; the lines in the boxes represent the medians; the whiskers represent non-outlier ranges; and the hollow dots represent outliers. (Del Gobbo et al., 2008). Asmentioned above, since the fish samples uti- lized by Del Gobbo et al. (2008) were obtained from grocery stores and the PFAS concentrations were relatively low (e.g., PFOS ranging 0.21– 1.68 ng/g ww) compared to those for sport fish, it is possible that ana- lytical uncertainty might have contributed to the proposed decline in the concentrations after the cooking. Concentrations of PFASs in the current study were approximately 10× higher than those in the sam- ples analyzed by Del Gobbo et al. (2008). We are unsure on the mecha- nism of loss during cooking; however, it is possible that the higher initial concentrations in this study obscured any small loss of PFASs that were noticeable in the Del Gobbo et al. (2008) study. In addition, someof the differences in the results from this study and those reported by Del Gobbo et al. (2008) can be attributed to differences in the cooking methods. For example, Del Gobbo et al. (2008) geared fish selection and cooking methods to include those common in Asian– Canadian populations. In the study by Del Gobbo et al. (2008), baked fish was marinated in rice wine, and fried fish was cooked in sesame oil 5 min longer than in the current study. Further, differences in the ratios of fish to oil and the surface areas of fish (e.g., small pieces versus larger portions affectingmoisture/fat loss)may have contributed to the differences in the results. The findings of this study are also in contrast to the observations for most neutral organic contaminants in fish, but are in agreement with the observations for some heavy metals (Bayen et al., 2005; Morgan et al., 1997; Perelló et al., 2008; Wilson et al., 1998). Neutral organic compounds aremainly associatedwith the fatty tissue of fish and are re- moved from fish fillet via dripping oil while cooking (Domingo, 2011). The heat applied to cook fish can also accelerate removal of neutral or- ganic compounds from the muscle via evaporation to air (Domingo, 2011). Therefore, even though some moisture in fish fillet is lost during cooking, the wet weight based concentrations of neutral organic com- pounds in fish fillet generally decrease after cooking. In contrast, PFASs and heavy metals are generally associated with proteins of biota and are less likely to be removed via cooking. As mass of fish fillet de- creases due to loss of moisture during cooking (Fig. 3), the wet weight based concentrations are likely to remain unchanged or increase after cooking. An increased wet weight based PFOS concentration after cooking, however, cannot singularly be interpreted as increased risk of human exposure to PFOS because exposure/risk depends on the total amount of PFOS in the fish (rather than concentration). Overall, changes in the fish PFOS amount after cooking (Fig. 4b, SI Table S4) differed signifi- cantly (p b 0.01) among fish species and cooking methods after allowing effects of each other; however, the interaction effect be- tween fish species and cooking methods was not significant (p = 0.25) (i.e., the effect of cooking on PFOS amount was not significantly affected by fish species). Declines in common carp PFOS amounts were significantly greater (p b 0.05) than those for Chinook salmon and walleye (Fig. 4b, SI Table S5). Among the cooking methods, dif- ferences existed between broiling and frying (Fig. 4b, SI Table S6). The amount of PFOS declined significantly (p b 0.05) in common carp after baking and frying, and increased significantly (p b 0.05) in lake trout after broiling (SI Tables S4 and S7). All other fish/cooking combinations showedno significant changes (p N 0.05) in PFOS amount (SI Tables S4 and S7). Although, frying seems to be more effective than baking and broiling in removing PFOS from the fish samples (SI Fig. S3), the reductions of PFOS amount were generally not substantial or statis- tically significant except for common carp (p N 0.05, SI Tables S4 and S7). The increases in fish PFOS amounts observed for a number of species/cooking combinations (Fig. 4b, SI Table S4) were also generally not significant (p N 0.05, SI Tables S4 and S7). Further, these increases (generally b20%) as well as wide ranges of change in PFOS amount for replicates (Fig. 4b) could be a result of ±10–15% uncertainty in both before and after cooking measurements. Similar to PFOS, other PFASs analyzed generally showed little statis- tically significant change in their amount after cooking (SI Tables S4 and 113S.P. Bhavsar et al. / Environment International 66 (2014) 107–114 S7). Chinook salmon fillets lost 14–20% of PFHxS and 25–28% of 6:8 PFPIA after cooking (SI Tables S4 and S7). Common carp also lost 4–20% of PFDA, PFUnDA, PFDoDA, PFTrDA, PFHxS, PFOS, 6:6 PFPIA and 6:8 PFPIA depending on the cooking method applied (SI Tables S4 and S7). Most of the statistically significant (p b 0.05) changes seen in lake trout PFAS amounts were minor (b15%) increases (SI Tables S4 and S7). Walleye lost 24 and 13% of 6:8 PFPIA when baked and fried, respectively (SI Tables S4 and S7). There are no certified referencematerials available for cooked fish to ideally test if recoveries from cooked fish are the same as from raw. However, recoveries of internal standard added to raw and cooked fish and seafood in Del Gobbo et al.(2008) did not differ (unpublished data, S. Tittlemier), indicating that the state of the fish (i.e., raw vs. cooked) does not affect fortified analyte recovery. In addition, Tittlemier et al. (2007) show that recoveries of PFASs for a variety of fortified food matrices including cooked chicken nuggets, organ meats, cured pork, and infant cereal were similar suggesting that there are no differences across these matrices. The studywas designed tominimize influence of potentially varying PFAS levels in different parts of the fillets on the study results (Fig. S1), cooked portions were not compared to identical position on the fish body for the raw portion. Although this could have contributed to the variability observed in the effects of the cookingmethods, no systematic differences in fish PFOS amounts after cooking across all the species in- dicate that the subsampling method did not have any major influence on the study results. 3.5. PFASs in blank canola oil and leftover juices All PFAS measurements in blank canola oil samples were below the detection limits (results not shown). Concentrations of PFASs in cooking juicesweremostly below the detection limits except for somemeasure- ments of PFOA (18% of the samples, ranged ND–0.03 ng/g ww), PFOS (93% of the samples, ND–1.3 ng/g), 6:2 diPAP (9% of the samples, ND– 0.09 ng/g ww), and 8:2 diPAPs (34% of the samples, ND–2.4 ng/g ww). PFOS concentrations and amounts (absolute value and as percentages of the amounts in the corresponding raw fish) in the juices collected after cooking are shown in SI Table S8. The amount of PFOS loss through the leftover cooking juices were b0.01% of the amount in the raw fish. Overall, the results show that PFAS losses to the cooking juices were negligible. 4. Conclusions A cooking experiment was conducted on skin-off fillets of Chinook salmon, common carp, lake trout and walleye collected from four rivers in Ontario, Canada to investigate the effectiveness of three cooking methods (baking, broiling, and frying) on reducing PFC levels in four fish species. In addition to PFASs such as PFCAs and PFSAs, PFASs of emerging concern PFPAs, PFPIAs and diPAPs were also considered. PFOS was the dominant PFAS detected and the concentrations were more than an order of magnitude higher than those collected from gro- cery stores in Canada, Spain, and China. This indicates that sport fish harvested near sites with a history of PFAS contamination could repre- sent a point source of dietary exposure to PFOS for recreational or subsis- tence fish consumers. Concentrations of PFOS in fish fillets generally increase after cooking; however, the amount of PFOS largely remains unchanged. Although changes in the PFAS amounts after cooking depended on fish species and cookingmethod used, cooking is generally not an effective approach to reduce dietary exposure to PFASs, especially PFOS. Conflict of interest The authors declare that they have no competing financial interests. Acknowledgments Funding for the study was provided by Health Canada through the Chemicals Management Plan Monitoring and Surveillance Fund. We thank Margaret Neff, Rachel Law, Scott Mabury and Miriam Diamond of the University of Toronto, and Chris Mahon, Justin Wilson and Dave Poirier of OMOE. Appendix A. Supplementary data Supplementary data (8 tables and 3 figures) to this article can be found online at http://dx.doi.org/10.1016/j.envint.2014.01.024. References Alhoniemi E, Himberg J, Parhankangas J, Vesanto J. SOM Toolbox 2.0. http://www.cis.hut. fi/projects/somtoolbox/about.shtml, 2005. [accessed Nov 2011]. Awad E, Zhang XM, Bhavsar SP, Petro S, Crozier PW, Reiner EJ, et al. Long-term environ- mental fate of perfluorinated compounds after accidental release at Toronto airport. Environ Sci Technol 2011;45:8081–9. Bayen S, Barlow P, Lee HK, Obbard JP. Effect of cooking on the loss of persistent organic pollutants from salmon. J Toxicol Environ Health A 2005;68:253–65. Begley TH, White K, Honigfort P, Twaroski ML, Neches R, Walker RA. Perfluorochemicals: potential sources of andmigration from food packaging. Food Addit Contam 2005;22: 1023–31. Bertelsen SL, Hoffman AD, Gallinat CA, Elonen CM, Nichols JW. Evaluation of log K-ow and tissue lipid content as predictors of chemical partitioning to fish tissues. Environ Toxicol Chem 1998;17:1447–55. Betts KS. Perfluoroalkyl acids — what is the evidence telling us? Environ Health Perspect 2007;115:A250–6. Bhavsar SP, Awad E, Mahon CG, Petro S. Great Lakes fish consumption advisories: is mer- cury a concern? Ecotoxicology 2011;20:1588–98. Canola Council of Canada. Fact sheet of canola oil. http://www.canolacouncil.org/media/ 501782/canola_oil.pdf. [accessed Mar 2013]. Conder JM, Hoke RA, DeWolfW, Russell MH, Buck RC. Are PFCAs bioaccumulative? A crit- ical review and comparison with regulatory lipophilic compounds. Environ Sci Technol 2008;42:995–1003. De Silva A, Gledhill M, Sekela M, Syrgiannis J, Evans M, Armellin A, et al. Perfluorinated chemicals in freshwater fish in Canada. Organohalogen Compd 2010;72:1277–80. Del Gobbo L, Tittlemier S, Diamond M, Pepper K, Tague B, Yeudall F, et al. Cooking decreases observed perfluorinated compound concentrations in fish. J Agric Food Chem 2008;56:7551–9. Delinsky AD, Strynar MJ, McCann PJ, Varns JL, McMillan L, Nakayama SF, et al. Geograph- ical distribution of perfluorinated compounds in fish fromMinnesota lakes and rivers. Environ Sci Technol 2010;44:2549–54. Domingo JL. Omega-3 fatty acids and the benefits of fish consumption: is all that glitters gold? Environ Int 2007;33:993–8. Domingo JL. Influence of cooking processes on the concentrations of toxic metals and var- ious organic environmental pollutants in food: a review of the published literature. Crit Rev Food Sci Nutr 2011;51:29–37. Dupont. Zonyl® and Foraperle®: Technical Information. http://www2.dupont.com/ Zonyl_Foraperle/en_US/tech_info/zonyl_techinfo.html, 2012. [Accessed Mar, 2012]. Egeland GM, Middaugh JP. Balancing fish consumption benefits with mercury exposure. Science 1997;278:1904–5. Ericson I, Gomez M, Nadal M, van Bavel B, Lindstrom G, Domingo JL. Perfluorinated chemicals in blood of residents in Catalonia (Spain) in relation to age and gender: a pilot study. Environ Int 2007;33:616–23. Ericson I, Marti-Cid R, Nadal M, Van Bavel B, Lindstrom G, Domingo JL. Human exposure to perfluorinated chemicals through the diet: intake of perfluorinated compounds in foods from the Catalan (Spain) Market. J Agric Food Chem 2008;56:1787–94. Giesy JP, Kannan K. Global distribution of perfluorooctane sulfonate in wildlife. Environ Sci Technol 2001;35:1339–42. Gokoglu N, Yerlikaya P, Cengiz E. Effects of cooking methods on the proximate composi- tion and mineral contents of rainbow trout (Oncorhynchus mykiss). Food Chem 2004;84:19–22. Gulkowska A, Jiang QT, So MK, Taniyasu S, Lam PKS, Yamashita N. Persistent perfluo- rinated acids in seafood collected from two cities of China. Environ Sci Technol 2006;40:3736–41. Guo R, Reiner EJ, Bhavsar SP, Helm PA, Mabury SA, Braekevelt E, et al. Determination of polyfluoroalkyl phosphoric acid diesters, perfluoroalkyl phosphonic acids, perfluoroalkyl phosphinic acids, perfluoroalkyl carboxylic acids, and perfluoroalkane sulfonic acids in lake trout from the Great Lakes region. Anal Bioanal Chem 2012;404: 2699–709. Hamza-Chaffai A, Cosson RP, Amiard-Triquet C, El Abed A. Physico-chemical forms of stor- age of metals (Cd, Cu and Zn) and metallothionein-like proteins in gills and liver of marine fish from the Tunisian coast: ecotoxicological consequences. Comp Biochem Physiol 1995;111C:329–41. Haukas M, Berger U, Hop H, Gulliksen B, Gabrielsen GW. Bioaccumulation of per- and polyfluorinated alkyl substances (PFAS) in selected species from the Barents Sea food web. Environ Pollut 2007;148:360–71. Health Canada. Canada's Food Guide. http://www.hc-sc.gc.ca/fn-an/food-guide-aliment/ index-eng.php, 2007. [accessed Oct 2011]. http://dx.doi.org/10.1016/j.envint.2014.01.024 http://www.cis.hut.fi/projects/somtoolbox/about.shtml http://www.cis.hut.fi/projects/somtoolbox/about.shtml http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0005 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0005 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0005 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0010 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0010 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0015 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0015 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0015 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0020 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0020 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0020 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0025 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0025 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0030 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0030 http://www.canolacouncil.org/media/501782/canola_oil.pdf http://www.canolacouncil.org/media/501782/canola_oil.pdf http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0035 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0035 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0035 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0040 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0040 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0045 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0045 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0045 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0050 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0050 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0050 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0055 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0055 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0115 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0115 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0115 http://www2.dupont.com/Zonyl_Foraperle/en_US/tech_info/zonyl_techinfo.html http://www2.dupont.com/Zonyl_Foraperle/en_US/tech_info/zonyl_techinfo.html http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0060 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0060 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0065 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0065 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0065 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0070 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0070 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0070 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0075 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0075 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0080 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0080 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0080 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0085 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0085 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0085 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0090 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0090 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0090 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0090 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0090 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0095 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0095 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0095 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0095 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0100 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0100 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0100 http://www.hc-sc.gc.ca/fn-an/food-guide-aliment/index-eng.php http://www.hc-sc.gc.ca/fn-an/food-guide-aliment/index-eng.php 114 S.P. Bhavsar et al. / Environment International 66 (2014) 107–114 Health Canada. Human health risk assessment of mercury in fish and health benefits of fish consumption. http://www.hc-sc.gc.ca/fn-an/pubs/mercur/merc_fish_poisson- eng.php, 2007. [Accessed Apr 2012]. Hori T, Nakagawa R, Tobiishi K, Iida T, Tsutsumi T, Sasaki K, et al. Effects of cooking on con- centrations of polychlorinated dibenzo-p-dioxins and related compounds in fish and meat. J Agric Food Chem 2005;53:8820–8. Houde M, De Silva AO, Muir DCG, Letcher RJ. Monitoring of perfluorinated compounds in aquatic biota: an updated review PFCs in aquatic biota. Environ Sci Technol 2011;45: 7962–73. Jones PD, Hu W, De CoenW, Newsted JL, Giesy JP. Binding of perfluorinated fatty acids to serum proteins. Environ Toxicol Chem 2003;22:2639–49. Kelly BC, Ikonomou MG, Blair JD, Surridge B, Hoover D, Grace R, et al. Perfluoroalkyl con- taminants in an Arctic marine food web: trophicmagnification and wildlife exposure. Environ Sci Technol 2009;43:4037–43. Kohonen T. Self-organizing maps. Berlin. New York: Springer; 2001. Lee H, Mabury SA. A pilot survey of legacy and current commercial fluorinated chemicals in human sera from United States donors in 2009. Environ Sci Technol 2011;45: 8067–74. LindstromAB, StrynarMJ, Libelo EL. Polyfluorinated compounds: past, present, and future. Environ Sci Technol 2011;45:7954–61. Mari M, Nadal M, Schuhmacher M, Domingo JL. Application of self-organizing maps for PCDD/F pattern recognition of environmental and biological samples to eval- uate the impact of a hazardous waste incinerator. Environ Sci Technol 2010;44: 3162–8. Martin JW, Smithwick MM, Braune BM, Hoekstra PF, Muir DCG, Mabury SA. Identification of long-chain perfluorinated acids in biota from the Canadian Arctic. Environ Sci Technol 2004;38:373–80. Mason Chemical Co. Masurf® FS-710 and 780: technical information. http://www. masonsurfactants.com/PDFs/Ind/Masurf_FS-710_and_FS-780_20101012.pdf, 2012. [Accessed Mar 2012]. Morgan JN, Berry MR, Graves RL. Effects of commonly used cooking practices on total mercury concentration in fish and their impact on exposure assessments. J Expo Anal Environ Epidemiol 1997;7:119–33. Mozaffarian D, Rimm EB. Fish intake, contaminants, and human health — evaluating the risks and the benefits. J Am Med Assoc 2006;296:1885–99. OMOE. Ontario Ministry of the Environment Guide to Eating Ontario Sport Fish. 26th ed.; 2011 [http://www.ene.gov.on.ca/stdprodconsume/groups/lr/@ene/@resources/ documents/resource/std01_079301.pdf (accessed Nov 2011)]. Perelló G, Marti-Cid R, Llobet JM, Domingo JL. Effects of various cooking processes on the concentrations of arsenic, cadmium, mercury, and lead in foods. J Agric Food Chem 2008;56:11262–9. Sherer RA, Price PS. The effect of cooking processes on PCB levels in edible fish tissue. Qual Assur 1993;2:396–407. Tittlemier SA, Pepper K, Seymour C, Moisey J, Bronson R, Cao XL, et al. Dietary exposure of Canadians to perfluorinated carboxylates and perfluorooctane sulfonate via con- sumption of meat, fish, fast foods, and food items prepared in their packaging. J Agric Food Chem 2007;55:3203–10. U.S. EPA. The national listing of fish advisories. http://water.epa.gov/scitech/swguidance/ fishshellfish/fishadvisories/advisories_index.cfm, 2010. [accessed Nov 2011]. UNEP. The nine new POPs. An introduction to the nine chemicals added to the Stockholm Convention by the Conference of the Parties at its fourth meeting. http://chm.pops. int/Programmes/NewPOPs/Publications/tabid/695/language/en-US/Default.aspx, 2009. [accessed Nov 2011]. Wania F. A global mass balance analysis of the source of perfluorocarboxylic acids in the Arctic Ocean. Environ Sci Technol 2007;41:4529–35. Weber J, Bochi VC, Ribeiro CP, Victorio AD, Emanuelli T. Effect of different cooking methods on the oxidation, proximate and fatty acid composition of silver catfish (Rhamdia quelen) fillets. Food Chem 2008;106:140–6. WHO. Diet, nutrition and the prevention of chronic diseases. JointWorld Health Organiza- tion and Food and Agriculture Organization of the UnitedNations expert consultation. Available online:http://www.fao.org/DOCREP/005/AC911E/ac911e00.htm, 2002. [Accessed Oct. 2011]. Wilson ND, Shear NM, Paustenbach DJ, Price PS. The effect of cooking practices on the concentration of DDT and PCB compounds in the edible tissue of fish. J Expo Anal En- viron Epidemiol 1998;8:423–40. Zhang T, Sun H, Lin Y, Wang L, Zhang X, Liu Y, et al. Perfluorinated compounds in human blood, water, edible freshwater fish, and seafood in China: daily intake and regional differences in human exposures. J Agric Food Chem 2011;59:11168–76. Zhang XM, Gandhi N, Bhavsar SP, Ho LS. Effects of skin removal on contaminant levels in salmon and trout filets. Sci Total Environ 2013;443:218–25. http://www.hc-sc.gc.ca/fn-an/pubs/mercur/merc_fish_poisson-eng.php http://www.hc-sc.gc.ca/fn-an/pubs/mercur/merc_fish_poisson-eng.php http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0135 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0135 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0135 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0140 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0140 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0140 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0145 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0145 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0150 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0150 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0150 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0155 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0160 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0160 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0160 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0165 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0165 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0170 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0170 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0170 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0170 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0175 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0175 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0175 http://www.masonsurfactants.com/PDFs/Ind/Masurf_FS-710_and_FS-780_20101012.pdf http://www.masonsurfactants.com/PDFs/Ind/Masurf_FS-710_and_FS-780_20101012.pdf http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0185 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0185 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0185 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0190 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0190 http://www.ene.gov.on.ca/stdprodconsume/groups/lr/@ene/@resources/documents/resource/std01_079301.pdf http://www.ene.gov.on.ca/stdprodconsume/groups/lr/@ene/@resources/documents/resource/std01_079301.pdf http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0200 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0200 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0200 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0205 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0205 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0210 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0210 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0210 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0210 http://water.epa.gov/scitech/swguidance/fishshellfish/fishadvisories/advisories_index.cfm http://water.epa.gov/scitech/swguidance/fishshellfish/fishadvisories/advisories_index.cfm http://chm.pops.int/Programmes/NewPOPs/Publications/tabid/695/language/en-US/Default.aspx http://chm.pops.int/Programmes/NewPOPs/Publications/tabid/695/language/en-US/Default.aspx http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0225 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0225 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0230 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0230 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0230 http://www.fao.org/DOCREP/005/AC911E/ac911e00.htm http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0240 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0240 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0240 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0245 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0245 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0245 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0250 http://refhub.elsevier.com/S0160-4120(14)00029-4/rf0250 Cooking fish is not effective in reducing exposure to perfluoroalkyl and polyfluoroalkyl substances 1. Introduction 2. Methods 2.1. Sample collection and preparation 2.2. Cooking details 2.2.1. Frying 2.2.2. Baking 2.2.3. Broiling 2.2.4. Post-cooking 2.2.5. Blanks 2.3. Sample preparation for PFAS analysis 2.4. Instrumental analysis 2.5. Data analysis 3. Results and discussion 3.1. Changes in sample mass after cooking 3.2. PFAS concentrations and profiles 3.3. PFAS patterns and clustering 3.4. Effect of cooking on PFASs in fish 3.5. PFASs in blank canola oil and leftover juices 4. Conclusions Conflict of interest Acknowledgments Appendix A. Supplementary data References