The common ingredients of the marinades were purchased at a local grocer. Frozen acerola puree (Bloch & Schou Aps, Viborg, Denmark) was freeze-dried before use. Oregano and sumac (ENES Gewürze GmbH, Geesthacht, Germany) were used dried as they were bought. The acerola and oregano marinade was an oil/water emulsion (O/W). The Dijon marinade and sumac marinade was oil based while the barbecue marinade was water-based. The ingredients are given in Table 1. Danish cold pressed rapeseed oil (Lehnsgaard, Aakirkeby, Denmark) was used as the base of the marinades.

Boneless pork loins (m. longissimus dorsi) with approximately 3 mm fat rim, purchased from a commercial Danish slaughterhouse, were used for all experiments. The loins were sliced into 15 chops of 2 cm thickness.

Two different consumer studies were performed: a central location test (CLT) of the appearance of the chops and a home-use test (HUT) of the three selected marinades to assess the liking after barbecuing.

The CLT was carried out in a canteen setting in Taastrup, a suburb of Copenhagen, the capital of Denmark. A total of 109 consumers participated (aged between 20 and 60 years, 61% male and 39% female). The consumers evaluated ten samples: the five marinades either as pre-marinated chops or presented next to the chops in a small bag (see Figure 1). The test was performed using a holistic by DMRI^TM approach (Jakobsen et al., 2014), in which the consumers evaluated the extent to which the following attributes were suitable for the product, on a 15 cm unstructured line scale anchored 1 cm from each end ranging from “not at all” (0) to “very much” (15): Familiar, Appetizing, Boring, Natural, Strange, Exciting, Sticky and Buying Intention. The order of the attributes was randomised between consumers except Buying Intention, which was always the last attribute. Samples were presented to the consumers in a randomised Latin square design.

Seventy-six families, recruited from the area around Copenhagen, participated in the HUT. Up to four persons per family could evaluate the samples, but from most families

two persons participated. A total of 176 consumers aged between 20 and 60 participated (95 males and 81 females). On a Friday, the consumers were given three packages of chops and the three marinades: acerola, Dijon and oregano, packed in small bags. The consumers were instructed to barbecue the chops at home on the same day during the weekend in the way they would normally do so. On a line scale (11.4 cm) ranging from “very bad” to “very good” anchored 1 cm from each end, the consumers assessed liking of the appearance and odour of the barbecued chop, followed by how much they liked the chop when eaten and finally whether they could imagine buying a chop with this marinade another time. The order of the chops was not controlled, but, since the consumers had to fill in the number of the marinades themselves, a random order between families was expected.

Each family member was asked to complete a short questionnaire including age in

10-year intervals, gender, how often they had barbecued chops in the barbecue season (every day, every week, every month, more seldom) and how often they ate marinated meat (every week, every month, more seldom).

On the Monday, the consumers returned the questionnaires together with one barbecued chop for each of the three types of marinade.

To investigate the effect of the marinades on reduction of the development of HCAs, controlled barbecue cooking was carried out indoor at Grilleriet (www.grilleriet.dk) with a controlled room temperature (20˚C) and a constant airflow and humidity. A Weber One-touch® premium barbecue and coconut charcoal (www.kokosbriketter, Videbæk, Denmark) were used. Both direct heat using a distance of 10 cm from the heat (open barbecue) and indirect heat after initially one minute of direct heat, with a distance of 1 cm from the heat on each side (closed barbecue), were used. When direct heat was used, the chops were turned every two minutes, and when indirect heat in the closed barbecue was used, the chops were turned every third minute to avoid having to open the lid too often. The temperature of the barbecue using direct heat was approximately 270˚C, measured with an infrared thermometer (BUHL & BØNSØE A/S, Virum, Denmark), while the temperature of the barbecue when using indirect heat was between 240˚C and 290˚C, measured with the barbecue’s built-in thermometer. Six chops with each marinade were grilled simultaneously on the same barbecue. Half of them were removed from the heat when they had reached a core temperature of 65˚C, while the remaining three were grilled until they had reached a core temperature of 80˚C. This was repeated twice, giving a total of six chops for each treatment (barbecue type and temperature). In addition to the three marinades, pork chops marinated in rapeseed oil and pork chops without any marinade (blank) were also grilled, giving a total of five different types (three marinades, rapeseed oil, blank), two barbecue types, two core temperatures and six replicates. The 30 treatments were arranged in a Latin square design on pairs of two loins.

Multispectral imaging (Videometer A/S, Hørsholm, Denmark) was used to estimate the core temperature. Based on the average surface spectra of the chops from the controlled cooking (open barbecue), three separate models were predicted by partial least squares regression (Unscrambler X10.3, Camo software A/S, Oslo, Norway)―one for each marinade. The core temperature of the HUT grilled pork chops was estimated from these models.

The HCAs were extracted from the meat crust using hollow-fibre (HF) extraction followed by an LC-MS/MS analysis [14] . In brief, prepared HF membranes were filled with 0.1 M sulfuric acid as an acidic acceptor solution and then immersed into 1-octanol to fill the fibre pores with the organic solvent.

Blended crust was homogenized in 0.05 M NaOH, and spiked with internal standards (harman-d3, norharman-d7, PhIP-d3), (100 μl, 2 mg/l). After mixing the homogenate for 1 h, followed by a centrifugation, 2 mL of the supernatant was transferred for extraction using the HF device. The HF was immersed into the sample and the extraction was carried out for 90 min. The sealed end of the HF was cut off and the acceptor phase inside the fibre lumen was transferred to a vial. The sample amount was adjusted to 0.100 g with Milli-Q water.

The chromatographic separations were performed with the following modifications: using an ACQuity UPLC BEH C18 (2.1 × 50 mm 1.7 µm) column at 40˚C (Waters Corporation, Milford, Mass., USA) and a Krud Katcher UHPLC in-line filter (0.5 µm depth filter x0.004 in ID, Phenomenex, Torrace, CA, USA) on a uHPLCN exera X2 system (Shimadzu, Kyoto, Japan). Separation was achieved with a binary mobile phase at a flow rate of 0.5 mL/min. Solvent A was acetonitrile, and solvent B was 10 mM ammonium formiate at pH 3.0. The gradient elution programme was: 0 - 0.2 min, 95% A, 0.2 - 6.0 10% A, 6.1 - 9.0 min 95% A, return to original condition. The injection volume was 5 µL. The analysis used a hybrid triple quadrupole (QqQ)/linear ion trap mass spectrometer from AB Sciex (QTRAP® 6500, AB SCIEX, Concord, ON, Canada).

The data from the CLT appearance test were analysed using Panel check (version 1.4.0, Nofima, Norway). A PCA analysis was performed using unstandardised data, first on all data followed by the subset of the chops presented with the marinade in a bag.

Liking data from the HUT were analysed using a mixed model analysis of variance with the marinade as fixed effect and the family as random effect (Proc mixed, SAS version 9.4, Cary NC, USA).

The logarithmic content of HCAs in the pork chops after controlled cooking was analysed using an analysis of variance with marinade, temperature, grill method and the interaction between these as fixed effects (Proc mixed, SAS version 9.4, Cary NC, USA). The post-hoc comparison was performed using Bonferroni adjustment for multiple comparisons.

In a preliminary study, the antioxidative of several herbs, berries and other active ingredients was assessed using two assays: oxygen consumption and DPPH. This was compared with literature data on ORAC values. On this basis, oregano, sumac and acerola was chosen for the further experiment. In the development of the marinades, other ingredients were added to get the best gastronomic quality. Some of these ingredients are also known to have antioxidative properties. A list can be seen in Table 2.

The main difference between the samples in the CLT of appearance was whether they were pre-marinated or packed with the marinade in a bag placed next to them (Figure 2, left plot). This accounted for 93% of the variation in the data (PC1). The pre-mari- nated samples were rated as boring, strange and sticky, while the chops with the marinade in the bag were rated as exciting, appetizing, natural, familiar and with a higher purchase intention score. It was therefore decided to continue the study with the marinade in bags instead of pre-marinated samples.

A closer examination of the samples with the marinades in the bag (right plot) showed that especially the acerola marinade was perceived as being strange and sticky, while especially oregano and Dijon were perceived as being appetizing, exciting, familiar and natural. Barbecue marinade was more sticky, familiar and natural compared

¹from a prestudy. ²μmol TE/100 g (micromol Trolox Equivalent per 100 grams). ³Scavenging activity in 5 mL DM/mL. ⁴from http://modernsurvivalblog.com/health/high-orac-value-antioxidant-foods-top-100/. ⁵from http://superfoodly.com/superfoodly-orac-values.pdf.

with sumac, which was perceived as being strange, but also exciting. Based on this, the oregano and Dijon marinades were chosen for the HUT and controlled cooking. However, since acerola has strong antioxidant properties, and since a closer examination of the data showed that a small segment of the consumers actually liked it (data not shown), this marinade was selected as the third in the HUT and controlled cooking.

In the HUT, the consumers assessed liking after barbecuing at home. In general, the consumers liked all three types of marinades, with all of them getting an average score above the middle of the scale (Table 3). However, for all attributes, the liking of the acerola marinade was lower than for the oregano and Dijon marinades, which were equally liked. In particular, the question “would you eat it again?” got low ratings for the acerola marinade.

The surface colour of the chops cooked by the consumers was measured using Videometer Lab and related to the surface colour of the chops from the controlled cooking in order to estimate the core temperature (Figure 3).

The meat was estimated to be cooked to below 65˚C by 42% of the consumers, between 65˚C and 70˚C by 24% and between 70˚C and 80˚C by 34%. The methodology of comparing the cooked surface of the consumer chops with the cooked surface of the chops from the controlled cooking is somewhat imprecise. Nevertheless, it was shown that in general only relatively few consumers (14%) cooked the chops to above 75˚C.

Three heterocyclic aromatic amines with mutagenic properties, MeIQx, DiMeIQx and PhiP, were analysed (Table 3(b)). Of these, PhiP was the most abundant. A three-way interaction between marinade, core temperature and cooking method existed for DiMeIQx and PhiP (P < 0.001) but not for MeIQx (P = 0.176), which had a two-way interaction between marinade and cooking method instead (P < 0.001). With indirect cooking, all of the marinades reduced the content of DiMeIQx at both temperatures,

¹Values < LOQ.

while the reduction was only significant for MeIQx in the oregano marinade, in which the formation of HCAs in general was almost totally inhibited. No effect was seen when cooking over direct heat for DiMeIQx and MeIQx. In contrast, PhiP was inhibited independently of the cooking method and temperature and was only insignificant for the oregano marinade when cooking over direct heat to a core temperature of 65˚C.

Harman and Norharman have comutagenic activity, meaning that they themselves have no mutagenic effects but can increase the mutagenicity of other HCAs (Jägerstad, Skog, Arvidsson, & Solyakov, 1998). Unfortunately, a consistent and very high concentration of, in particular, Harman, but also Norharman, was seen after marinating with the Dijon marinade. The concentration was much higher than in the control sample. No clear reducing effect of the other marinades was seen on Harman and Norharman compared with the blank sample.

Cold pressed rapeseed oil used for the marinades has an antioxidant effect on its own, and it was therefore of interest also to investigate the effect of the pure oil. No significant reduction of pure rapeseed oil was seen for any of the investigated HCAs except Norharman cooked over direct heat to 80˚C. However, a numerical decrease was seen, indicating that the oil might have some properties that inhibit the formation of HCAs.

Barbecuing is one of the cooking methods that results in the highest content of HCAs [26] , and previous research has shown that approximately half of the consumers in a home use study used marinades before barbecuing pork chops [14] . This provides an opportunity to develop tailor-made marinades that inhibit the formation of HCAs based on the knowledge about the antioxidant activities of different herbs and berries. All three marinades consisted of several ingredients, as the aim was to investigate the effect of a complex marinade. Other ingredients than acerola, sumac and oregano has therefore contributed to the antioxidative effect of the marinade. However, the aim was not especially to focus on the effect of these three ingredients, but to demonstrate the possibility of combining gastronomy and chemical research to develop effective marinades, including ingredients that initially have a high antioxidant effect. Success will depend on the sensory quality of the marinades being high as well. The sensory quality of the marinated meat has only been included in a few studies, in which more or less trained panels were used [8] [27] [28] . In this study, consumer tests were used to assess the sensory quality. In comparison, other studies have focused on marinades that did not alter the sensory quality [28] . In the CLT, the consumers clearly appreciated the marinade being presented in a bag together with the pork chops instead of pre-mari- nated chops.

Based on the test of the appearance, three marinades were chosen for further study― Dijon and oregano, which were clearly appreciated, and acerola, due to its promising antioxidant activity, even though only a few of the consumers liked the appearance. This lower liking was also found during the HUT, with the acerola marinade getting lower scores for all attributes, especially the question “would you eat it again?”, compared with the Dijon and oregano marinades.

When the controlled grilled samples were used to estimate the core temperature of the consumer samples, it was seen that only approximately one third of the consumers cooked the meat to above 70˚C. This is in contrast to a previous study in Denmark, in which more than half of the estimated core temperatures in consumer-cooked pork chops were 75˚C or more [14] . This difference might be due to the fact that the relationship between well-done meat and cancer has received media coverage in Denmark since the first study. This is important, since longer cooking times increase the concentration of, in particular, PhiP and MeIQx, but also diMeIQx [1] , which was also confirmed in this study, where the content of all HCAs was higher in the chops cooked to 80˚C compared with 65˚C in the control samples. Shorter cooking times resulting in a lower degree of doneness are therefore recommended.

In the controlled barbecuing, coconut charcoal was used because they are said to give a higher and more stable temperature (Grilleriet®, personal communication), which is crucial to keeping the conditions as controlled as possible during the cooking process. It has been shown that the concentration of PhiP is lower in meat cooked on this type of charcoal compared with wood charcoal [29] . Other types of charcoal might therefore have resulted in a higher formation of HCAs and thereby also a greater potential of reducing PhiP.

PhiP was the most abundant HCA, which confirms several other studies [1] [12] [30] [31] , although MeIQx, DiMeIQx, Harman and Norharman were also present in detectable concentrations. The reducing effect of the three marinades was not consistent for all five HCAs analysed in this study. For MeIQx and DiMeIQx, a reduction was generally seen for all marinades when barbecuing on indirect heat, whereas no effect was seen when barbecuing on direct heat. The concentration of PhiP was also reduced by the marinades for both types of barbecue at both temperatures. In contrast, the concentration of Harman, and to a lesser extent Norharman, was increased considerably by the Dijon marinade, even though the antioxidant potential of this marinade was similar to that of the other marinades. The Dijon marinade contained oregano, sumac and acerola as well as Dijon mustard, brown sugar and soy sauce. The increasing effect of Dijon marinade on Harman and Norharman was not seen in the other two marinades, which also contained oregano and acerola, and it must therefore have been caused by some of the other ingredients. Previously, it has been indicated, that sugar as well as soy sauce might increase the formation of MeIQx, 4, 8DiMeIQx and PhiP [32] . This was not seen in this study, but it cannot be rejected, that the sugar and soy sauce are responsible for the increase in Harman and Norharman. The Dijon marinade cannot be recommended due to its high concentrations of Harman and Norharman, even though the content of the other mutagenic HCAs is very low.

The acerola marinade was neither well-liked nor as effective as the oregano marinade in reducing the HCAs, especially at 65˚C, even though acerola is known to be an effective antioxidant [21] . The reason for this lower efficiency is not known, although it could be due to the different compositions of the active groups in oregano, which are a varity of polyphenols, flavonoids and different compounds in the essential oil such as carvacrol, thymol, γ-terpinene and linalool [17] [18] and acerola, which are a combination of ascorbic acid [22] and phenolic compounds [23] . Another explanation could be the other ingrediens in the oregano marinade as especially garlic and onions are both known to have a high antioxidative capacity (Table 2) and a high efficiency in reducing the formation of HCAs [33] , but also thyme, which is present in a lower concentration (Table 1) can reduce the formation of HCA’s though to a lesser extend than oregano [34] . This leaves the oregano marinade as the most appreciated and most effective marinade for healthy barbecuing.