Exposure to acrylamide from home-cooked food: fried potatoes (rösti) in Switzerland as an example

ABSTRACT

There is a significant lack of data on acrylamide in food which is prepared domestically, as obtaining samples of food cooked in people’s homes is more complex than taking samples from shops. Rösti is a fried potato dish, which can be a significant contributor to acrylamide exposure, particularly in the German speaking areas of Switzerland. A set of 71 samples was collected from people who cooked the dish at home. The average acrylamide content was 709 µg/kg. Based on a food frequency questionnaire and on a Swiss national food survey the results enabled the determination of the exposure of the population to acrylamide through home-made rösti. The values for an exposure estimate to acrylamide from rösti are 7.8 µg/day for the whole population (including rösti eaters and non-eaters) and 27.2 µg/day for predominantly rösti consumers participating in this study.

KEYWORDS:

• Acrylamide
• reducing sugar
• fried potatoes
• monitoring
• exposure
• risk
• rösti
• hash browns
• home-cooking
• domestic cooking

Introduction

It was a surprise when researchers discovered in 2002 that acrylamide, which is a common industrial chemical, can be formed during food processing at high temperatures (Tareke et al. 2002). The biggest contribution to this process contaminant stems from Maillard reactions of asparagine and reducing sugars such as fructose and glucose (Mottram et al. 2002; Stadler et al. 2002). Frying potatoes to make potato crisps, hash browns or French fries provides favourable conditions for the formation of acrylamide, as potatoes contain high levels of asparagine and can contain high levels of reducing sugars.

Since 2002 the toxicological properties, exposure and risk of acrylamide have been extensively studied. Risk assessments generally conclude that the dietary exposure of the population to acrylamide is of concern, the latest EFSA opinion from 2015 is no exception (EFSA CONTAM Panel 2015). The margins of exposure (MOE) for the mean exposure estimates vary between 1433 and 226 for neurotoxic effects and 567 and 89 for neoplastic effects. EFSA concluded in earlier opinions that, for substances which are both genotoxic and carcinogenic, an MOE of 10000 or more can be considered of low concern from a public health point of view (EFSA 2005b; EFSA SC 2012a). It seems undisputed that acrylamide exposure of consumers is of concern.

EFSA estimates the dietary exposure of the 95^th percentile across all age groups and studies to be between 0.6 and 3.4 µg/kg b.w. per day (EFSA CONTAM Panel 2015). The exposure estimates were based on extensive occurrence data containing about 46 000 data points, which were obtained by EU member states and food-industry associations. Closer inspection of the data, however, reveals that most of the data was obtained from ready to eat products – foremost potato crisps, which made up 34 000 of the data points. The data on fried potato products were mostly obtained from French fries. In fact, there were only 96 results from “other fried potato products”, 12 of these were rösti.

Though there is a substantial amount of information available on mitigation methods for acrylamide formation in food (FoodDrinkEurope 2019; U.S. Food and Drug administration (FDA) (2016), unfortunately, there has been limited improvement for consumers with respect to acrylamide dietary exposure in the last decade. Furthermore, EFSA had not been mandated to evaluate the effectiveness of mitigation measures in their recent opinions (EFSA CONTAM Panel 2015). Such measures may include rules on potatoes used for making French fries. In Switzerland French fries are traditionally made from carefully chosen potatoes and therefore contain less acrylamide than internationally standardised French fries from fast food chains (Biedermann et al. 2010). Another mitigation measure implemented in Switzerland by the potato distributing industry is the labelling of potatoes as suitable for frying and baking which are available in supermarkets in the months following harvest. These potatoes are stored at 8–9 °C in order to prevent them from producing large amounts of reducing sugar, something that happens readily at the conventional storage temperature of 4 °C (Noti et al. 2003). An intended side effect of the campaign described here, was a promotion of the potatoes stored at higher temperature. The EU has recently introduced legislation, in an attempt to reduce acrylamide exposure. The impact on the population’s dietary exposure to acrylamide remains to be seen. The recent Commission Regulation (EU) 2017/2158 of 20 November 2017 “establishing mitigation measures and benchmark levels for reducing the presence of acrylamide in food” basically makes detailed industry measures a legal requirement for food business operators. The concept of the legislation considers periodic adjustment to the benchmark levels (90^th percentile), as acrylamide contents in various food categories decrease. So far there has been no adjustment.

Rösti is a popular Swiss dish not unlike American hash browns. It consists of grated and fried potatoes. Some consumers eat rösti once or several times a week. For the German speaking Swiss rösti consuming population, it can be the biggest source of dietary exposure to acrylamide (Grob 2005). One study reported that a portion (250 g) of rösti can contain 1000 µg of acrylamide (Biedermann 2010), which would lead to an exposure of 1 µg/kg b.w. per day, for a consumer of 60 kg, who eats such a rösti once every two weeks – and that is a potential exposure from rösti alone. As stated above the mean total dietary acrylamide exposure of the population according to EFSA is also in the range of 1 µg/kg b.w. per day. This makes rösti an ideal target to reduce the dietary acrylamide exposure in the population. Although rösti is a regional speciality, the results are transferrable to other fried potato dishes, which are common all over Europe and in large parts of the world.

Exposure assessments of the population to acrylamide frequently fail to adequately address the exposure through home cooking. This is due to the circumstance that most measurements are done by authorities or industry on ready to eat products (EFSA CONTAM Panel 2015). For instance, the Austrian food safety authority, in their acrylamide monitoring from 2007–2015, concluded that acrylamide exposure from domestic cooking or restaurants was not considered (Lückl et al. 2016). They suggest a systematic investigation of acrylamide content in food from people’s homes and restaurants (Lückl 2016). In 2015 a study reported the results of 55 röstis from restaurants in Zurich (McCombie et al. 2016). The average acrylamide content was 702 µg/kg and 16 % higher than the value given by EFSA for “other fried potato products”. The restaurants had been invited to submit a rösti for a free analysis of the acrylamide content, and were, therefore, aware of the goal of the analysis (McCombie et al. 2016).

In this study, 71 röstis were obtained directly from people’s homes, without informing the participants of what was being analysed, thus providing important additional information in the exposure assessment. Additionally, the participants filled in a food frequency questionnaire (FFQ) with regards to their consumption of rösti and other fried potato products. Together with the data from the national food survey ‘menuCH’ (Chatelan et al. 2017a; 2017b; 2018; SFVO 2019) this enables the determination of the whole population’s exposure.

Materials and methods

Samples

In the local version (for Zurich) of a free national commuter’s newspaper a short article was placed asking people to provide the official food control laboratory of the canton of Zurich with home-made rösti, prepared from whole potatoes (no pre-packed, convenience rösti) in April 2016. Interested people were referred to a short URL with instructions on how to proceed. The readership of the Zurich version of the publication was estimated to be 0.5 million people. Participants were asked to sign up (enabling the authors to limit the number of samples), to prepare at least one portion of rösti (250 g), wrap it into tin foil, freeze it and bring it to the laboratory. The participants were not informed on the purpose of the analysis when the samples were provided, in order to prevent any bias. During sign up, gender, age, address and type of preparation were queried. Participants were paid CHF 25.- for preparing and bringing in a sample and were, in addition, asked to complete an FFQ to establish their eating habits. From the people who signed up (n = 97) a few dropped out, resulting in 71 samples. All participants received a test report with their own results as well as the anonymised results of all the other participants. The participants consisted of 32 women and 39 men; 17 were aged 17 to 30, 39 were aged 31 to 60 and 15 over 60; 31 lived in the town of Zurich, at least 27 prepared their rösti by frying uncooked potatoes (reported as “from uncooked potatoes”). Rösti can be made by frying raw or (partially) boiled potatoes.

Acrylamide

“Acrylamide was analysed according to a previously published method (Biedermann et al. 2002). In brief, the samples were soaked in water at 70 °C for 30 min and extracted with 1-propanol. The solvent was evaporated and the residue partitioned between hexane and acetonitrile. The latter was analysed using on-column injection and gas chromatography-mass spectrometry (GC-MS) with chemical ionisation (CI). The uncertainty of the results was below 15% for contents above 100 µg/kg and below 25% for lower contents.” (McCombie et al. 2016) Extraction was performed on a visually representative and large section of 70 g (more than a quarter) of the total rösti to ensure the variability introduced would be smaller than the uncertainty of the measurement.

Fructose and glucose

Glucose and fructose were determined enzymatically with Boehringer Mannheim test kits purchased from R-Biopharm AG (Darmstadt, Germany; article number 10139106035) as described in a previous publication (Mini et al. 2004; McCombie et al. 2016). Briefly, 40 g rösti was homogenised with 250 ml distilled water using a Polytron (Kinematica, Lucerne, Switzerland). Then, 5 ml of each solution Carrez I (potassium hexacyanoferrate(II) trihydrate) and Carrez II (zinc sulphate heptahydrate) were added. In addition, 2 ml 1-octanol (Fluka) was added to prevent foaming and the volume adjusted to 500 or 250 ml with distilled water depending on the amount of potato. The samples were filtered (Schleicher & Schuell) or centrifuged at 5000 rpm and subjected to enzymatic analysis as described by the kit provider. The test kit provider gives an uncertainty of measurement of 20 % for each of fructose and glucose.

Colour index

“The colour index was calculated from photographs of the röstis before analysis. Photographs were imported into Igor Pro (Wavemetrics, Lake Oswego, OR). With the aid of an in-house macro the region of interest was chosen (the area of the photograph in which the rösti could be seen) and the average sum of all three colour layers (RGB – red/green/blue) was calculated, resulting in a low value for dark areas and high values for light areas. As the lighting was not identical on all pictures the value was normalised to a section of white on each photograph which contained the label with the sample number. The actual colour index value is the inverse of the normalised average sum of red, green and blue, making a high index value darker than a low one.”(McCombie et al. 2016) The uncertainty for the colour index was not determined, but is expected to be small due to its definition and normalisation to the white label.

Food frequency questionnaire (FFQ)

All participants were asked to fill in a FFQ. They were asked how often (between never and twice or more a day in 9 steps) and what size portions (with pictures to illustrate one portion) they eat rösti, fried potatoes, roasted potatoes, chips (French fries) and potato crisps. The uncertainty of the FFQ was not estimated. The self-reported values by the participants were taken at face value.

General consumption data for the Swiss population

The consumption data on the general population for rösti was calculated from menuCH, the Nutrition Survey in Switzerland. menuCH is a nationwide cross-sectional survey conducted in ten study centres among a stratified random sample of 2057 adults aged 18 to 75 years from January 2014 to February 2015 (Chatelan et al. 2017b). The dietary assessment is based on two non-consecutive 24-hour dietary recalls. The first dietary recall was face-to-face and the second by phone, two to six weeks later. To support participants in quantifying consumed amounts, a book with series of graduated portion-size pictures and a set of actual household measures were additionally used (Chatelan et al. 2017b).

Results and discussion

As potatoes contain similar amounts of asparagine independent of variety or storage this parameter cannot be influenced by a cook (Armein et al. 2003). Two other parameters with the potential to increase acrylamide formation, however, can be influenced by the chef: reducing sugars and temperature of frying (Biedermann-Brem et al. 2003; Fiselier et al. 2006). Therefore, to make a tasty rösti with a low acrylamide content, potatoes with low contents of fructose and glucose must be used as well as a low frying temperature applied. These, therefore, are the parameters which were analysed in the samples in this study.

In this study, 71 samples of rösti, prepared by members of the public in their homes, were analysed for their content of fructose, glucose and acrylamide. These measurements enabled assessment of the cook’s frying process and his or her choice of potatoes used. An optimally tasty rösti should be crispy with a hint of caramel (which is formed in part through Maillard reactions) and be “golden” in colour, without excessive browning. Black rösti develops a bitter taste. Potatoes high on reducing sugars have the potential to brown quickly and form high levels of acrylamide. An attentive chef, who erroneously uses potatoes with too much fructose and glucose would have to abort his frying process due to excessive browning. It was demonstrated by McCombie et al. (2016) with the reference röstis that aborting the frying process by an extra attentive chef who deliberately used potatoes which were not ideal for making rösti lead to around 700 µg/kg acrylamide in these products (Figure 1).

Figure 1. Summary of all the results of acrylamide and reducing sugar contents from this study (black circles). For comparison to professionally prepared samples the data for restaurant rösti (grey circles) and reference rösti (boxes) were also added to the graph (McCombie et al. 2016)

The contents of acrylamide and reducing sugars varied considerably across all the samples. The acrylamide content ranged from 25 to 2930 µg/kg (see Table 1 for results). Figure 1 gives a graphic representation of the results (black circles) where the fructose + glucose content is plotted against the acrylamide content. This allows a comparison with professionally prepared (grey circles) and reference röstis (boxes) from McCombie et al. (2016). The “professionally prepared” röstis were cooked by chefs at restaurants, who were offered a free analysis, of acrylamide in their rösti, if they were prepared to provide one. The reference röstis had been prepared by the Belvoirpark School of Hotel Management Zurich. The chef used potatoes with varying contents of reducing sugars and prepared a culinarily acceptable product while keeping the browning and therefore acrylamide content to a minimum.

Table 1. Measured levels of acrylamide and reducing sugars glucose and fructose (gluc + fruct) in 71 homemade röstis ordered by acrylamide level

Sample#	Acrylamide[ug/kg]	gluc + fruct[g/kg]	Sample#	Acrylamide[ug/kg]	gluc + fruct[g/kg]
2	25	15.8	52	600	15.3
28	40	9	39	610	9.7
27	45	0.57	4	640	5.4
45	50	10.7	60	650	12.9
38	120	9.3	55	665	10.7
31	130	4.95	46	710	12.8
16	150	6.5	17	715	3.6
56	190	3.75	44	720	12.6
71	195	3.52	51	725	5.3
37	200	16.2	40	740	14.5
68	210	9.3	69	835	7.9
26	225	5.4	36	855	20.2
15	235	3.08	10	860	13.8
21	240	1.73	67	860	12
53	250	1.08	64	880	4.8
3	280	12.4	47	895	13
22	280	9.3	33	900	12.9
9	290	20	63	940	13.3
50	300	1.55	7	945	7.8
41	355	2.86	65	975	17.9
30	375	2.89	19	1020	19.2
8	380	25	25	1020	11
20	425	6.8	29	1130	21.2
62	435	2.38	12	1170	5.4
14	440	11.1	48	1240	6.5
13	445	9.5	11	1250	12.9
61	460	14.1	70	1250	17.8
59	485	1.79	54	1260	15.1
32	490	12.9	49	1300	8.2
58	530	14.8	18	1490	12.1
43	535	15.6	1	1510	17.5
66	535	6.8	23	1620	8.2
35	570	15.2	24	1640	14.1
6	575	15.1	5	1830	25
42	575	10.8	57	1910	6.2
			34	2930	10.2

It can clearly be seen, that the influence of the frying process is large as there is no correlation between the reducing sugar and acrylamide contents for the data of this study. The mean acrylamide content in this study was 709 µg/kg. As this is a high value and a portion of rösti is quite large, it is worth thinking about how this can relate to exposure: the consumption of an average rösti every third day results in an acrylamide exposure of 1 µg/kg b.w./day from rösti alone. This value is in the range of the total mean dietary exposure to acrylamide across all age groups and surveys in the most recent EFSA opinion on acrylamide (EFSA CONTAM Panel 2015). The average acrylamide content is also 16 % higher than the value EFSA gives for “other fried potato products” which include rösti. The reference röstis contained a maximum of 710 µg/kg acrylamide irrespective of sugar content, indicating that all the participants whose rösti contain more acrylamide than this value, could optimise their frying.

All attempts failed to find a correlation between any of the parameters which were collected about the participants and the acrylamide content. For instance the average acrylamide content for men (n = 39) was 711 µg/kg and for women (n = 32) 707 µg/kg. The average content for frying raw potatoes was 701 µg/kg, which does not support the hypothesis that frying raw potatoes may lead the cook to allow more acrylamide formation. This had been hypothesised in the (smaller) dataset of restaurant röstis, as using raw potatoes requires a longer frying time (McCombie et al. 2016).

The rösti’s content of reducing sugars (the sum of fructose and glucose) also varied greatly between samples ranging from 0.6 to 25 g/kg. Though there are several post-harvest parameters, which impact the reducing sugar accumulation in potatoes (Kumar et al. 2004), the biggest influence is from storage temperatures. Potatoes intended for frying should be stored at 8–9°C, instead of the more common 4°C to prevent sugar formation (Noti et al. 2003). The storage of potatoes leads to a clear seasonal variation in potatoes’ sugar contents and potential for acrylamide formation, as freshly harvest potatoes do not contain high amounts of fructose and glucose (Powers et al. 2013). The study was run at a time of year (April), where only 5 röstis contained less than 2 g/kg of reducing sugars, as it was a long time after harvest time and potatoes have usually been stored for a long time. Practically all potatoes sold in Swiss supermarkets and which would be used for preparing rösti are Swiss potatoes harvested in autumn. Cooks who want to prepare good, low-acrylamide rösti need to use low sugar potatoes. The reference rösti with a sugar content of 1.2 g/kg contained only 190 µg/kg acrylamide. The dataset clearly shows there is very little awareness among the participants, with respect to the use of optimal potatoes for an optimal culinary result, which would also mitigate acrylamide formation.

When comparing the data of home-made rösti in this study with the previous restaurant study (McCombie et al. 2016) there are two notable differences. The first (and minor one) is that there are more low sugar, high acrylamide rösti prepared by restaurants. The authors could not find a very satisfactory explanation for this phenomenon. Possibly, the professional chefs insist more on browning and, therefore, fry for too long and/or too hot, while people at home may tend to cook to time. The other difference is due to the sugar contents. The range of reducing sugars determined in the home-made röstis extends to much higher values than for the restaurant rösti. For the restaurant samples the reducing sugars were measured in a sample of potatoes before frying, while for the home-made samples they were determined in the rösti itself. This difference, however, does not explain the high sugar values for home-made rösti, as a rösti only loses about 10 % of its weight (water) during frying (data not shown). We therefore concluded that the potatoes for the 10 samples with sugar contents over 16 g/kg must have been stored in the refrigerator by the participants. It is plausible that potatoes are stored for longer in people’s homes than in restaurants. In future studies, it would worth considering to query the source of the potatoes and storage conditions. However, this would have to be done after the rösti is prepared to prevent giving the participants hints as to the relevant parameters.

Generally, the variation of culinary quality in the home-made rösti seems bigger by visual inspection. For instance, sample 2 with an acrylamide content of 25 µg/kg and 15.8 g/kg reducing sugars was basically grated boiled potatoes. There was no trace of frying, though the sample had the characteristic shape it gets from a frying pan.

The browning of rösti (and other fried potato products) is due to the Maillard reaction and can be used as a good indicator for acrylamide formation (Mestdagh et al. 2008).

Using image analysis on the photographs of the samples an index for the colour can be calculated. A high index value indicates a “dark” rösti. Figure 2 shows the colour indices for each sample plotted against its acrylamide content and reveals a clear correlation even though neither the thickness of the rösti or the dark layer (“dilution”) nor the back side were considered. The samples from the restaurant study (McCombie et al. 2016) were added to the plot for comparison.

Figure 2. Summary of all the colour indices from this study (black circles). For comparison to professionally prepared samples the data for restaurant rösti (grey circles) and reference rösti (boxes) were also added to the graph (McCombie et al. 2016)

The curve flattens somewhat at higher acrylamide levels as the thickness of the dark layer increases (and the colour index remains high). If samples are mixed during frying, the colour index can remain low at relatively high acrylamide levels. Despite the scatter in the data, the plot demonstrates that checking the rösti for browning is an effective way for a cook to monitor the formation of acrylamide. In the past the correlation of browning and acrylamide content has been used for the “Swiss frying test” (Mini et al. 2004). In this test potatoes were heated at standardised conditions and the browning permitted the selection of suitable potatoes with a low potential for acrylamide formation. Four samples in this study stand out with high acrylamide content in the range of 1500 µg/kg but low colour index values. This was most likely due to the cooks mixing the rösti during frying, which is sometimes done.

All participants were required to fill in a food frequency questionnaire, in which they were asked how often they consume five different potato based foods. They also had to estimate the amount they ate each time. The results of the frequency questions are summarised in Figure 3.

Figure 3. Answers to food frequency questionnaire in this study. The frequency of consumption of five different potato based foods were queried. Tick options were never, one time per month (1 x/m), 2–3 times per month (2–3 x/m), one time per week (1 x/w), two times per week (2 x/w), 3–4 times per week (3–4 x/w), 5–6 times per week (5–6 x/w), daily (1 x/d) and more than once daily (>1 x/d)

All but four of the participants (5.6 %) were rösti consumers who most commonly eat rösti between one and three times per month (n = 59; 83 %). Together with the estimated portion sizes a daily rösti intake could be calculated for each participant. The 50^th percentile (P50) for rösti intake for the participants in this study was 16.7 g/day. When multiplied with the 95^th percentile of the acrylamide content of all röstis in this study, which was 1630 µg/kg this leads to an exposure of 27.2 µg acrylamide per day and person. More exposure scenarios are given in Table 2 and compared to the exposure scenarios for the entire population including non-consumers of rösti derived from the menuCH data.

Table 2. Exposure calculations of the Swiss population to Acrylamide from rösti (*mean of 4114 interviews with 116 reported rösti consumptions)

	P50 consumption	P95 consumption	P50 acrylamide content	P95 acrylamide content	P50 consumption xP50 acrylamide content	P50 consumption xP95 acrylamide content	P95 consumption xP50 acrylamide content	P95 consumption xP95 acrylamide content
	[g/day]		[µg/kg]		Exposure [µg/day]
This Study	16.7	57	600	1630	10	27.2	34.2	93
menuCH data	4.8*	-			2.9	7.8	-	-

In the menuCH study, of the 2057 representative participants who reported their food consumptions on two days six weeks apart 116 counts of rösti consumption were reported. From the study design it is not possible to calculate the number of non-consumers. If a rösti consumer would consume rösti once a month, the probability of that consumer eating rösti on one of the two interview days would be a mere 1/15^th. The data show how much rösti is consumed on the 4114 days by a representative proportion of the population, i.e. the mean consumption of rösti including consumers and non-consumers. In the menuCH study, the mean value for the entire population was 4.8 g rösti per day. Using the acrylamide data (P50 and P95) from this study, this leads to an acrylamide exposure of 2.9 resp. 7.8 µg per day and person. Unsurprisingly, the inclusion of all non-consumers using the menuCH data leads to lower exposure estimates. For comparison and context: In EFSA’s 2015 report the total mean dietary acrylamide exposure was given as 24–114 µg per day and 60 kg bodyweight (EFSA CONTAM Panel 2015). This indicates that, using the acrylamide values of this study and the consumption data of the entire Swiss population, rösti is a large contributor to acrylamide exposure and one which might have been underestimated by EFSA as there are so few available data on home cooking. Rösti consumers have an exposure from rösti alone in the range of the average total exposure. Rösti and other home-made fried potato products, therefore, remain an effective target for reducing the overall acrylamide exposure of the population.

Conclusions

EFSA’s 2015 risk assessment for dietary exposure to acrylamide (EFSA CONTAM Panel 2015) may not adequately consider the exposure to acrylamide from home cooking and restaurant prepared foods (except French fries). In this study 71 röstis prepared in the homes of members of the public were analysed. The participants were not informed what the goal of the study was in advance (but were provided an analysis report after completion of the measurements). The average acrylamide content was 709 µg/kg. This value is 16 % higher than the 606 µg/kg stated in the aforementioned EFSA opinion. There was no difference in the average acrylamide content in rösti made by men or women, in town or in the countryside or between rösti made from raw or boiled potatoes.

The contents of acrylamide and reducing sugars were very similar to those obtained from a similar study of restaurant röstis (McCombie et al. 2016). One difference is that potatoes with low reducing sugar contents were more likely to be prepared so they have less acrylamide by members of the public. Possibly this is because professional chefs may use excessive heat to achieve browning despite potatoes low on reducing sugars. The other notable difference is that the sugar content of some samples was very high. It was hypothesised that some participants store their potatoes in their fridges.

The exposure of the participants to acrylamide was 27.2 µg/day and person from rösti alone. This is a large proportion of the estimated dietary acrylamide exposure by EFSA of the whole population. This means that EFSA’s exposure assessment most likely fails to include exposure from home cooking adequately, especially for high consumers of fried potatoes. Further investigations into home cooking and possible mitigation measures for public health need to be addressed in the future.

Acknowledgments

The authors would like to thank the Food Safety and Veterinary Office (FSVO) for funding this project and Vincent Dudler from the FSVO for his support.

The authors also thank the IT-team and reception of the Official Food Control authority of the Canton of Zurich for their support with the project.

Lastly, many thanks go to the 71 individuals who were prepared to cook a rösti they were not able to eat.