Risk presented to minimally processed chilled foods by psychrotrophic Bacillus cereus

Abstract

Background

Spores of psychrotrophic Bacillus cereus may survive the mild heat treatments given to minimally processed chilled foods. Subsequent germination and cell multiplication during refrigerated storage may lead to bacterial concentrations that are hazardous to health.

Scope and approach

This review is concerned with the characterisation of factors that prevent psychrotrophic B. cereus reaching hazardous concentrations in minimally processed chilled foods and associated foodborne illness. A risk assessment framework is used to quantify the risk associated with B. cereus and minimally processed chilled foods.

Key findings and conclusions

Bacillus cereus is responsible for two types of food poisoning, diarrhoeal (an infection) and emetic (an intoxication); however, no reported outbreaks of food poisoning have been associated with B. cereus and correctly stored commercially-produced minimally processed chilled foods. In the UK alone, more than 10¹⁰ packs of these foods have been sold in recent years without reported illness, thus the risk presented is very low. Further quantification of the risk is merited, and this requires additional data. The lack of association between diarrhoeal food poisoning and correctly stored commercially-produced minimally processed chilled foods indicates that an infectious dose has not been reached. This may reflect low pathogenicity of psychrotrophic strains. The lack of reported association of psychrotrophic B. cereus with emetic illness and correctly stored commercially-produced minimally processed chilled foods indicates that a toxic dose of the emetic toxin has not been formed. Laboratory studies show that strains form very small quantities of emetic toxin at chilled temperatures.

Highlights

• •Psychrotrophic B. cereus is a hazard to safety of minimally processed chilled foods.
• •Spores may survive food processing, with bacterial growth during chilled storage.
• •Illness not reported for B. cereus and correctly stored processed chilled foods.
• •Lack of foodborne illness probably due to low pathogenicity at chilled temperature.
• •High quality materials/manufacturing standards, and chill chain contribute to safety.

1. Introduction

Minimally processed chilled foods have organoleptic characteristics that satisfy consumer demand for high quality foods of high nutritional value and easy preparation. These foods are often of near neutral pH and with little/no added salt or preservative, and rely on a combination of good quality raw materials, mild heat treatment (generally at 70 °C–90 °C), manufacturing hygiene, chilled storage and a restricted shelf-life to ensure product safety and quality (Carlin et al., 2000; Peck, 2006; Peck, Goodburn, Betts, & Stringer, 2008). Chilled storage in the UK, except Scotland, is specified as 8 °C or below (FSA, 2016), with different temperatures specified in some other countries. Minimally processed chilled foods represent an ever-growing share of the diet in many European countries. For example, the estimated total value of the UK chilled prepared food market in 2018 was ca. £12bn (Chilled Food Association, 2018).

The principal food safety hazard to these foods is endospore-forming bacteria that survive the minimal heat process and multiply at chilled temperature (Carlin et al., 2000). The risk presented to these products by non-proteolytic Clostridium botulinum has been widely documented (Barker, Malakar, Plowman, & Peck, 2016; Lindström, Kiviniemi, & Korkeala, 2006; Malakar, Barker, & Peck, 2011; Peck, 2006, 2009). Spores of this pathogen can survive the mild heat treatment applied, with spore germination potentially followed by cell multiplication and formation of the highly potent botulinum neurotoxin at a temperature as low as 3.0 °C (Carter and Peck 2015; Graham, Mason, Maxwell, & Peck, 1997; Peck, 2009; Peck, Stringer, & Carter, 2011). Consumption of even small amounts of pre-formed botulinum neurotoxin in food can lead to the severe disease, botulism, and potentially death.

The Gram-positive, facultatively anaerobic bacterium Bacillus cereus is ubiquitous in the environment and can be recovered from food and food materials. It also forms heat resistant spores that are likely to survive the mild heat treatments given to these foods. Some B. cereus strains can multiply at 8 °C and below to a concentration that may be detrimental to human health, and therefore represent a hazard to the consumption of minimally processed chilled foods. Although the various toxins formed by B. cereus are not as potent as the neurotoxins formed by C. botulinum, they can lead to serious forms of foodborne illness. Bacillus cereus is associated with two types of food poisoning: diarrhoeal and emetic. The former is associated with consumption of B. cereus followed by growth of the organism and de novo enterotoxin formation in the GI tract, and the latter with consumption of food containing pre-formed cereulide toxin produced by B. cereus during growth in food (Table 1). Foods previously associated with diarrhoeal food poisoning include meat products, fish, poultry, soups, sauces and stews, milk and milk products and vegetables (Carlin & Nguyen-The, 2013; Ceuppens et al., 2011; EFSA, 2016; Stenfors Arnesen, Fagerlund, & Granum, 2008). Emetic food poisoning is associated with consumption of a toxic dose of cereulide toxin that has been pre-formed by B. cereus in farinaceous foods, such as rice, pasta, noodles, potatoes, bread and pastries (Agata, Ohta, & Yokoyama, 2002; Ceuppens et al., 2011; Ehling-Schulz, Frenzel, & Gohar, 2015).

Table 1.

The two types of food poisoning caused by Bacillus cereus.

	Diarrhoeal syndrome	Emetic syndrome
Toxins	Nonhaemolytic enterotoxin (Nhe) Haemolysin BL (Hbl) Cytotoxin K (CytK)	Cereulide (Ces)
Dose necessary for illness	≥ 105 cfu g−1 of food a	~400 μg cereulide b
Requirements for illness	Growth in food to an infectious dose, consumption of which leads to infection and formation of toxins in small intestine of host	Cereulide production in food at high cell concentration. Illness caused by ingestion of food containing pre-formed toxin (intoxication)
Incubation time	8–24 h	0.5–5 h
Duration of illness	12–24 h	6–24 h
Toxin produced	Small intestine of host	Pre-formed in foods
Toxin properties	Heat labile proteins (inactivated by 56 °C/5 min) Inactivated by proteases and pH < 4.0	Heat stable cyclic peptide (no loss of activity at 121 °C for 90 min) Not inactivated by protease enzymes Stable at pH 2 to 11
Foods implicated	Meat products, fish, poultry, soups, sauces and stews, milk products and vegetables	Farinaceous foods such as rice, pasta, noodles, potatoes, bread and pastries

Adapted from Ceuppens et al. (2011), Ceuppens et al. (2013), Stenfors Arnesen et al. (2008).

^aconcentration unacceptable for ready-to-eat foods (EFSA, 2016; Health Protection Agency, 2009).

^bquantity of cereulide necessary to cause illness to a 50 kg human assuming 8 μg cereulide kg⁻¹ is the toxic dose for humans (Jääskeläinen et al., 2003).

Bacillus cereus sensu lato (the B. cereus group) consists of several closely related species, including B. anthracis, B. cereus sensu stricto, B. cytotoxicus, B. mycoides, B. pseudomycoides, B. thuringiensis, B. toyonensis, B. weihenstephanensis and B. weidmannii. However, it is not always possible to distinguish between different species in the B. cereus group using traditional phenotypic and biochemical characteristics (Kovac et al., 2016), and several classification systems have been proposed based on molecular typing methods (Guinebretière et al., 2008, 2010) and whole genome sequencing (Böhm, Huptas, Krey, & Scherer, 2015; Kovac et al., 2016; Liu et al., 2015). These suggest that species names alone are not adequate to describe the taxonomy of the B. cereus group. Guinebretière et al. (2008) separated the B. cereus group into seven phylogenetic groups, irrespective of species designation, and included data on phenotypic characteristics of the strains in each group, such as temperature growth ranges. This classification system has been adopted in this review as mesophilic and psychrotrophic strains cluster into different phylogenetic groups (Guinebretière et al., 2008, 2010), which is useful when considering the hazard posed by B. cereus in minimally processed chilled foods.

Phylogenetic groups I, III, IV and VII contain mesophilic or moderately thermotolerant strains, and are unable to multiply at 8 °C (Table 2). Strains responsible for causing foodborne illness typically belong to phylogenetic group III or IV. Strains in phylogenetic groups II, V and VI are psychrotrophic and grow at, and in many cases also below 8 °C (Table 2) and are the focus of this review as they may be able to grow during storage of minimally processed chilled foods to a level sufficient to cause illness. Strains of psychrotrophic B. cereus are less frequently involved in foodborne illness than mesophilic strains. For example, Glasset et al. (2016) characterised 339 B. cereus strains isolated from 74 foodborne outbreaks in France between 2007 and 2014 where B. cereus was the only pathogen isolated and found that 70% of isolates belonged to the mesophilic groups III and IV, with fewer isolates belonging to psychrotrophic group II (19%), group V (2%) or group VI (4%). Importantly, although strains of psychrotrophic B. cereus groups II, V and VI have been isolated from foodborne outbreaks, previous reports have failed to identify food poisoning outbreaks associated with correctly stored commercially-produced chilled foods (Daelman, Jacxsens, Devlieghere, & Uyttendaele, 2013; EFSA, 2005; Kennedy, 2004). An analysis of 81 food poisoning outbreaks linked to the consumption of cooked chilled foods in Australia and New Zealand implicated B. cereus in ten outbreaks (Kennedy, 2004). However, correctly stored commercially-produced minimally processed chilled foods were not implicated in any of these outbreaks; the main causative factors were inadequate cooling, storage and/or reheating. An EFSA Opinion stated that refrigerated foods had seldom been linked to outbreaks of Bacillus food poisoning (EFSA, 2005), and Daelman, Jacxsens, Devlieghere, et al. (2013) referred to the microbial safety of these foods as satisfactory. To update these reports, a literature survey was conducted using Web of Science (All Databases, 1945 to November 2018 with the following search terms: cereus, and food poisoning, and either minimally processed chilled food or chilled food or ‘ready meal’ or REPFED or cook chill or cooked chilled). This literature survey failed to identify outbreaks of food poisoning associated with B. cereus and correctly stored commercially-produced chilled foods.

Table 2.

Properties of the Bacillus cereus group.

Phylogenetic Group	Species	Temperature growth range (°C)	Growth limits:			Spore heat resistance		Pathogenicity potential (in cytotoxicity assays)
			NaCl (%)	aw	pH	Median time to first ten-fold reduction at 90 °C (min)	D90°C-value (min)
I	B. mycoides B. pseudomycoides	10 to 43	5	0.965	4.6	NR	NR	Low/none
II	B. cereus B. thuringiensis B. weidmannii	6 to 43	8	0.945	≤4.3	30	5 to 19	Moderate
III	B. cereus B. anthracis B. thuringiensis	13 to 48	≥10	≤0.929	≤4.3	39	10 to 102	High
IV	B. cereus B. thuringiensis	10 to 48	≥10	≤0.929	4.6	28	7 to 52	High
V	B. cereus B. thuringiensis B. toyonensis	8 to 43	8	0.945	4.6	20	10 to 45	Moderate
VI	B. cereus B. mycoides B. thuringiensis B. weihenstephanensis	5 to 40	6	0.960	4.6	2	1 to 13	Low/none
VII	B. cytotoxicus	18 to 56	>10	≤0.929	≤4.3	90	25 to 101	Very high

Adapted from Afchain et al. (2008), Carlin et al. (2013), Guinebretière et al. (2008), Guinebretière et al. (2010), Luu-Thi et al. (2014), Techer et al. (2014).

NR = not reported.

Based on the sales of commercially-produced minimally processed chilled foods in the UK (estimated as more than 10¹⁰ packs between 1995-2005 (Peck et al., 2008), with no reported cases of food poisoning associated with correctly stored chilled food and B. cereus, then the risk presented by these products is likely to be low. However, it should be noted that foodborne illness caused by B. cereus is usually mild and self-limiting (Bennett, Walsh, & Gould, 2013), so the true incidence of food poisoning due to B. cereus is likely to be under-reported (Adak, Long, & O'Brien, 2002; Scallan et al., 2011; Thomas et al., 2015). Details of food poisoning outbreaks involving B. cereus have been summarised previously (Gormley et al., 2011; EFSA & ECDC, 2016; Bennett et al., 2013; Santé Publique France, 2017), and occasionally diarrhoeal and emetic food poisoning have led to hospitalisation and death (Dierick et al., 2005; Lund, De Buyser, & Granum, 2000; Mahler et al., 1997; Naranjo et al., 2011; Saleh et al., 2012).

This review considers the risk presented by strains of B. cereus to commercially-produced minimally processed chilled foods. For this review, minimally processed chilled foods are considered as those heated at 70 °C–90 °C, and then stored at 8 °C or below. “Psychrotrophic Bacillus cereus” is defined as a B. cereus strain or phylogenetic group that can grow at a temperature of 8 °C or below.

2. Properties of Bacillus cereus

2.1. Presence of Bacillus cereus in the environment and foods

Spores of B. cereus are ubiquitous in the environment and can be recovered from soil at concentrations as high as 10⁶ cfu g⁻¹ (Hendriksen, Hansen, & Johansen, 2006; von Stetten, Mayr, & Scherer, 1999). Food and food materials can become contaminated with spores of B. cereus (Table 3). A survey of 51,165 samples of raw materials used in the preparation of UK commercially-produced minimally processed chilled foods and the foods themselves revealed low B. cereus contamination. For 99.5% of samples tested the B. cereus concentration was <10² cfu g⁻¹, 0.4% of samples contained 10²-10³ cfu g⁻¹, <0.1% of samples contained 10³-10⁴ cfu g⁻¹, and no samples had a B. cereus load greater than 10⁴ cfu g⁻¹ (Table 3). An analysis of Belgian data collected in 2009 provided an insight into minimally processed chilled foods produced by five companies (Daelman, Jacxsens, Devlieghere, et al., 2013). For 95.9% of samples tested on the day of production, and 97.9% of samples tested at the end of shelf-life, the B. cereus concentration was <10² cfu g⁻¹. The B. cereus concentration exceeded 10³ cfu g⁻¹ in 0.5% of samples tested on the day of production, and none exceeded 10³ cfu g⁻¹ at the end of shelf-life (Daelman, Jacxsens, Devlieghere, et al., 2013). Similar loadings of B. cereus have been found in other surveys of foods sampled from manufacturing, retail, and catering premises (Table 3). A B. cereus concentration of greater than 10⁵ cfu g⁻¹ is deemed unsatisfactory for ready-to-eat foods in the UK (Health Protection Agency, 2009), and the same or similar limits are applied in other EU countries (EFSA, 2016). Guidelines for B. cereus and other pathogenic Bacillus spp. in Australia/New Zealand for ready-to-eat foods sampled at the point of consumer sale/distribution describe four categories of microbiological quality: (i) < 10² cfu g⁻¹, satisfactory; (ii) 10²–10³ cfu g⁻¹, marginal but acceptable microbiological quality; (iii) 10³–10⁴ cfu g⁻¹, unsatisfactory and outside acceptable microbiological limits; (iv) > 10⁴ cfu g⁻¹, potentially hazardous (FSANZ, 2018). Additionally, it should be noted that not all B. cereus isolates are able to grow at ≤8 °C. A literature survey (comprising 32 studies) of the ability of B. cereus strains to grow at chill temperature indicates approximately 40% of the several thousand B. cereus strains examined in these studies can grow at ≤8 °C (Supplementary Table 1). Mesophilic strains were isolated more frequently than psychrotrophic strains in surveys recently carried out in Tunisia and South Africa (Gdoura-Ben Amor et al., 2018; Mugadza & Buys, 2017).

Table 3.

Examples of Bacillus cereus load of foods materials and foods sampled at the manufacturing, retail or catering level.

Country	Sampling period	Food products tested	Number tested	Percentage B. cereus load (cfu g−1)			Reference
Australia	2006–2007	Ready-to-eat/assemble foods	755	98.8 (<102)	1.2 (103–104)		Eglezos, Huang, Dykes, and Fegan (2010)
		Ready-to-re-heat foods	360	96.7 (<102)	3.3 (103–104)
		Raw foods	148	97.3 (<102)	0.7 (103–104)	2.0 (>104)
Australia	2016–2017	Extended shelf-life cook chill foods	98	96 (<102)	3 (102–103)	0 (103–104)	Aietan (2017)
						1 (104–105)
Belgium	2009	Commercial Refrigerated Processed Foods of Extended Durability	1299	97.4 (<102)	2.6 (102–103)	0.2 (>104)	Daelman, Jacxsens, Devlieghere, et al. (2013)
Denmark	2000–2003	Fresh food	991	97.2 (<103)	2.6 (103–104)	0.2 (>104)	Rosenquist, Smidt, Andersen, Jensen, and Wilcks (2005)
		Heated food	30,581	99.0 (<103)	0.6 (103–104)	0.4 (>104)
		Combined raw and heat-treated food	13,873	98.7 (<103)	0.7 (103–104)	0.6 (>104)
England	1997–1999	Ready-to-eat foods		99.8 (<104)	0.2 (>104)		Willis and Greenwood (2003)
England	2013	Ready-to-eat meat pies	862	98.8 (<103)	1.0 (103–105)	0.2 (≥105)	McLauchlin et al. (2016)
England	2012–2013	Liver pâté	870	99.0 (<103)	0.9 (103–105)	0.1 (≥105)	McLauchlin et al. (2017)
Germany	2008–2009	Pork + mustard/beer-marinade Pork + paprika marinade Pork + herb/garlic-marinade	100 100 100	87 (<101) 77 (<101) 72 (<101)	10 (101–102) 20 (101–102) 23 (101–102)	3 (102–103) 3 (102–103) 5 (102–103)	Pichner et al. (2014)
Netherlands	2002–2003	Oil(s) and fat(s) and products	60	100 (<105)			Wijnands, Dufrenne, Rombouts, in't Veld, and Van Leusden (2006)
		Fish and fish products	102	100 (<105)
		Meat and meat products	280	100 (<105)
		Flavourings	1384	99.9 (<105)	0.1 (≥105)
		Milk and milk products	5943	99.8 (<105)	0.2 (≥105)
		Ready-to-eat foods	22,744	99.7 (<105)	0.3 (≥105)
		Vegetable(s) and vegetable products	637	99.8 (<105)	0.2 (≥105)
		Pastry	2637	>99.9 (<105)	0.04 (≥105)
Netherlands	2007–2010	Various food materials	849	94.7 (<102)	2.6 (102–103) <0.1(104-105)	2.0 (103–104) <0.1 (>105)	Biesta-Peters et al. (2016)
Tunisia	2014–2015	Various food materials	687	77 (<103)	16 (103–104)	7 (>104)	Gdoura-Ben Amor et al. (2018)
Wales	1995–2003	Ready-to-eat foods	15,228	99.9 (<104)	0.08 (>104)		Meldrum et al. (2005)
Wales	2003–2005	Ready-to-eat foods	3391	99.9 (<104)	0.12 (>104)		Meldrum, Smith, Ellis, and Garside (2006)
UK	2015–2016	Raw materials and minimally processed chilled foods	51,165	99.5 (<102)	0.4 (102–103)	<0.1 (103-104) 0 (>104)	Chilled Food Association, unpublished data

2.2. Spore heat resistance

Bacillus cereus produces endospores, which are characterised by metabolic dormancy and resistance to extreme conditions such as high acidity and temperature, desiccation, UV irradiation, and exposure to chemicals and enzymes. Spores formed by strains of psychrotrophic B. cereus phylogenetic groups II and V are unlikely to be greatly affected by heat treatments (typically at 70 °C–90 °C) applied to minimally processed chilled foods (Table 2). However, spores formed by psychrotrophic B. cereus phylogenetic group VI strains are more heat sensitive than those of the other B. cereus groups (Table 2) and may be affected by some of the higher heat treatments used. A median time to the first ten-fold reduction in spore concentration of 2 min was reported at 90 °C, and the D_90°C-value for ten group VI strains ranged from 1 to 13 min (Afchain, Carlin, Nguyen-The, & Albert, 2008; Luu-Thi, Khadka, & Michiels, 2014). Deviations from first-order kinetics have been observed in thermal death curves for B. cereus spores (Carlin et al., 2006; Kort et al., 2005; Kramer & Gilbert, 1989). Consequently, heat resistance of B. cereus spores may be underestimated, which could affect the risk presented to minimally processed chilled foods by this organism. Additionally, spores of psychrotrophic B. cereus strains may be damaged/inactivated in the production of heavily heat processed ingredients (e.g. white powders such as starch), and thus be less prevalent in these raw materials than spores of more heat resistant mesophilic B. cereus strains (Malakar, Barker, & Peck, 2004).

2.3. Growth of Bacillus cereus strains at 8 °C and below

To have the potential to cause a food poisoning incident with correctly stored minimally processed chilled foods, psychrotrophic B. cereus strains need to grow to a hazardous concentration at refrigeration temperatures. The ability to grow at ≤8 °C is restricted to B. cereus phylogenetic groups II, V and VI (Table 2), and there is variability in the capability of B. cereus strains from these three groups to grow at refrigeration temperatures. Samapundo, Heyndrickx, Xhaferi, and Devlieghere (2011) determined the minimum growth temperature for 380 B. cereus strains isolated from food materials in Belgium, and found that the majority of the isolates did not grow at ≤8 °C: 88% of the isolates grew at 10 °C, 50% at 9 °C, 7% at 8 °C, 3% at 7 °C, <1% at 6 °C, and none at 5 °C. Under otherwise optimum conditions, approximately 80% of group II strains tested by Guinebretière et al. (2008), Luu-Thi et al. (2014) and Techer et al. (2014) could grow at 8 °C, as did 15% of group V strains, and 100% of group VI strains. No group II nor group V strains grew at 5 °C, but approximately 40% of group VI strains did. These percentages are likely to be lower at non-optimum pH/water activity. Growth of strains from other B. cereus groups may be an issue following temperature abuse. For example, although unable to grow at ≤8 °C, all group IV strains tested by Guinebretière et al. (2008) and Luu-Thi et al. (2014) grew at 10 °C.

2.4. Effect of other environmental factors on growth of Bacillus cereus

We surveyed the pH and water activity of 38 minimally processed chilled foods obtained from retail in the UK. Based on the reported growth limits for the B. cereus phylogenetic groups (Table 2), product pH (range 4.8–7.7; mean 6.0 ± 0.5) and water activity (range 0.962–0.996; mean 0.99 ± 0.01) are unlikely to prevent growth of psychrotrophic B. cereus. Although, B. cereus is a facultative anaerobe, growth in anaerobic conditions is often associated with longer lag times and slower growth rates, weak sporulation, and reduced biomass formation, particularly when combined with chilled incubation or increased acidity (Abbas, Planchon, Jobin, & Schmitt, 2014; de Sarrau et al., 2012; Guérin, Dargaignaratz, Broussolle, Clavel, & Nguyen-the, 2016; Samapundo, Everaert, et al., 2011; Thorsen, Budde, Koch, & Klingberg, 2009). A carbon dioxide concentration of ≥40% is reported to prevent growth of B. cereus at ≤8 °C (Bennik, Smid, Rombouts, & Gorris, 1995; Samapundo, Everaert, et al., 2011). On their own, individual controlling factors are unlikely to prevent growth of psychrotrophic B. cereus in correctly stored commercially-produced minimally processed chilled foods. However, combinations of anaerobiosis (and/or elevated CO₂ concentration) and non-optimal environmental conditions may be effective at limiting growth, particularly in combination with a heat treatment and chilled storage (Daelman, Sharma, et al., 2013; Daelman, Vermeulen, et al., 2013; Guérin et al., 2016; Samapundo, Heyndrickx, Xhaferi, de Baenst, & Devlieghere, 2014).

2.5. Formation of toxins by strains of Bacillus cereus

The B. cereus group form a range of toxins and virulence factors which may cause pathogenicity in humans (Castiaux, Laloux, Schneider, & Mahillon, 2016; Jeβberger et al., 2015; Kovac et al., 2016). Three enterotoxins; nonhaemolytic enterotoxin (Nhe), haemolysin BL (Hbl) and cytotoxin K (CytK) are likely to be responsible for causing diarrhoeal food poisoning (Table 1). All strains of the B. cereus group (including psychrotrophic B. cereus) appear to possess genes required to produce at least one of these diarrhoeal enterotoxins, but the extent of pathogenicity is influenced by factors such as the number and type of diarrhoeal enterotoxin genes present, their expression, and importantly the quantity of enterotoxin formed (Böhm et al., 2015; Castiaux et al., 2016; Glasset et al., 2016; Guinebretiere, Broussolle, & Nguyen-The, 2002; Guinebretière et al., 2010; Hendriksen et al., 2006; Jeβberger et al., 2015; Kovac et al., 2016; Miller, Jian, Beno, Wiedmann, & Kovac, 2018; Samapundo, Heyndrickx, et al., 2011; Stenfors, Mayr, Scherer, & Granum, 2002), and not all strains may form sufficient of the diarrhoeal enterotoxin(s) to cause foodborne illness (Miller et al., 2018; Stenfors Arnesen et al., 2008). The possession of the ces gene and the ability to form the emetic toxin (cereulide) is limited to a fraction of strains in the B. cereus group, and details are emerging of the regulation of, and environmental cues affecting emetic toxin formation, for example toxin formation is not favoured by anaerobiosis or chilled temperature (Biesta-Peters, Dissel, Reij, Zwietering, & in't Veld, 2016; Ehling-Schulz et al., 2015; Guérin, Thorsen Rønning et al., 2017). Several isoforms of cereulide have been identified, with varying potency (Marxen et al., 2015).

Strains of mesophilic groups III and IV are most commonly associated with foodborne illness, and are highly toxic, while psychrotrophic groups II and V contain strains that are moderately toxic, and group VI strains are of low toxicity (Table 2).

3. Consideration of previous exposure assessments and risk assessments

3.1. European food safety opinions

The European Food Safety Authority (EFSA) evaluated the risk to consumers of food poisoning due to B. cereus and other Bacillus species (EFSA, 2005, 2016). They considered that B. cereus-mediated food poisoning was commonly linked to heat-treated foods, and that B. cereus would be present in many types of food products at low concentrations (usually too low to represent a risk to consumers). To prevent growth of all known B. cereus strains in foods, EFSA deemed it necessary to have pH < 4.5, water activity <0.92 and storage temperatures <4 °C. However, EFSA acknowledged that temperatures less than 10 °C significantly increased lag and generation times of B. cereus, especially when other conditions in foods were sub-optimal for growth.

Based on the observations that, (i) the onset of symptoms of diarrhoeal food poisoning occur 6–24 h after food consumption, (ii) the mean transit time through the small intestine is estimated to be 4 h, and (iii) enterotoxin production occurs during the exponential growth phase, EFSA concluded that psychrotrophic B. cereus strains are less pathogenic and less likely to cause diarrhoeal illness than mesophilic food poisoning strains; as they have lower growth rates at 37 °C and form little or no enterotoxin at 37 °C and have low or no cytotoxic activity at 37 °C (EFSA, 2005, 2016). The EFSA opinion also recognised that the mechanisms leading to diarrhoeal food poisoning are very complex and not fully understood and stated that “refrigerated foods have seldom been incriminated in Bacillus foodborne outbreaks” (EFSA, 2005). Due to variability in pathogenic potential of B. cereus group strains, it is difficult to define the minimum infectious dose required to cause diarrhoeal illness in humans, and the current EFSA opinion recognises the need for further information (EFSA, 2016).

Regarding the ability of psychrotrophic B. cereus group strains to cause emetic food poisoning, the EFSA (2005) opinion states that these strains “do not seem able to produce emetic toxin and would only pose a low food poisoning risk”. However, this opinion was published prior to the discovery of a limited number of psychrotrophic strains that form small quantities of cereulide at chill temperatures. The recent opinion (EFSA, 2016) acknowledges that emetic food poisoning may have been associated with B. cereus concentrations as low as 10³ cfu g⁻¹; and recognises the need to identify the minimum intoxication dose for cereulide in humans as the current estimates of 8 μg kg⁻¹ body weight are based on animal studies.

3.2. Published exposure assessments for Bacillus cereus in minimally processed chilled foods

Several exposure assessments have been published for psychrotrophic B. cereus and minimally processed chilled foods (e.g. Afchain et al., 2008; Daelman, Membre, et al., 2013; De Cesare, Vitali, Trevisani, Bovo, & Manfreda, 2017; Malakar et al., 2004; Nauta, Litman, Barker, & Carlin, 2003; Rigaux, Ancelet, Carlin, Nguyen-thé, & Albert, 2013). The outcome of example assessments for two food products are now considered.

Daelman, Membre, et al. (2013) established a probabilistic model for exposure assessment in relation to hazards associated with psychrotrophic B. cereus in minimally processed chilled foods, with a model three component food considered rather than an actual product (Daelman, Membre, et al., 2013). A process risk model with nine modules was used to assess the process from raw materials to heated product, through chilled storage to consumption. A combination of data (new and literature), predictive models and expert opinion was used to build/model each module. The nine modules were; (i) raw material loading, (ii) cross contamination during handling and preparation, (iii) thermal inactivation (90 °C for 20 min), (iv) growth during intermediate storage (up to one day), (v) partitioning of batches, (vi) mixing to create the final product, (vii) recontamination during assembly and packaging, (viii) thermal inactivation during in-pack pasteurisation (90 °C for 10 min), and (ix) growth during chilled storage for up to 27 days. It was estimated that B. cereus was present at the point of consumption in 46.8% of packs (generally at a low concentration), with 0.5% of packs containing ≥10⁵ cfu g⁻¹. The assessment did not discriminate between diarrhoeal and emetic illness. Four key points were identified as important in limiting the exposure; (i) raw material contamination, (ii) recontamination during packaging, (iii) thermal death during cooking and pasteurisation, and (iv) most importantly effective refrigerated storage at retail and by the consumer (Daelman, Membre, et al., 2013). The model indicates the potential for exceeding a B. cereus concentration of 10⁵ cfu g⁻¹ of approximately once in every 200 retail units. Although, this level of exposure is greater than might be expected, based on an extensive literature survey described earlier that failed to link B. cereus associated illness and correctly stored minimally processed chilled foods, it should be noted that temperature abuse was a critical factor in this high level of exposure: 84% of the packs calculated to contain ≥10⁵ cfu g⁻¹ B. cereus at the point of consumption came from simulations where the consumer refrigerator temperature was >8 °C (Daelman, Membre, et al., 2013). Although the model construction is very detailed there are several areas where the representation of spore properties includes substantial uncertainty so that the outcomes cannot easily be applied to other minimally processed chilled foods prepared using different heat treatments or storage regimes. It is noted that the product considered by Daelman, Membre, et al. (2013) received a higher heat treatment and had a longer shelf-life than a majority of minimally processed chilled foods considered in the present study.

Afchain et al. (2008) published a quantitative exposure assessment to describe the transmission of B. cereus along the food pathway for a minimally processed chilled courgette purée that took account of the phylogenetic diversity of B. cereus. The spore load distributions for six of the B. cereus phylogenetic groups (group I was excluded) were estimated at each step in a processing chain comprising; raw materials, cooking (90 °C for 161 min), blending, mixing with uncooked ingredients providing potential secondary contamination, partitioning into 400 g retail packs, pasteurisation in a sealed vacuum pack (90 °C for 4 min), and chilled storage for up to 21 days. The model predicted that at the end of chilled storage, most packs will contain spores of B. cereus, and that spores of the mesophilic groups will be more prevalent than those of the psychrotrophic groups. Spores of psychrotrophic group II were predicted to be more widespread than spores of groups V and VI. The low incidence of group V was attributed to low contamination of raw materials, and the low incidence of group VI to low spore heat resistance (Afchain et al., 2008).

An extended second order assessment for the minimally processed chilled courgette purée was published by Rigaux et al. (2013). The quantitative assessment, built on the first order model of Afchain et al., included additional experimental data. Table 4 shows the new (posterior) predicted percentage of, (i) units containing at least one spore (for each phylogenetic group) at key stages of the food pathway, and, (ii) the percentage of each B. cereus phylogenetic group in proportion to the total B. cereus load at key stages of the food pathway. The model predicted that at the end of chilled storage, spores of mesophilic groups will be more prevalent than those of psychrotrophic groups, with spores of psychrotrophic group II more widespread than spores of psychrotrophic groups V and VI (Table 4). Given that most minimally processed chilled foods receive a considerably lower heat treatment and have shorter shelf-life than that described by Afchain et al. (2008), it is difficult to compare this outcome with those for other food processes.

Table 4.

Quantitative exposure assessment model for prevalence of spores and fractional loads segmented by Bacillus cereus group during production of a minimally processed chilled courgette purée (adapted from Afchain et al. (2008) and Rigaux et al. (2013)).

B. cereus group	Mean percentage of units containing at least one spore (percentage of B. cereus phylogenetic group relative to total B. cereus load)
	Raw courgettes	Cooked courgettesa	Mixing and partitioning into 400 g packsb	After pasteurisationc	End of chilled home storaged
IIe	100 (43)	64 (48)	29 (4)	13 (5)	13 (10)
III	100 (0.3)	48 (9)	72 (13)	33 (10)	33 (11)
IV	100 (0.4)	50 (7)	100 (83)	89 (85)	89 (79)
Ve	100 (0.4)	27 (0.2)	4 (0.2)	0.1 (<0.1)	0.1 (<0.1)
VIe	100 (56)	57 (22)	9 (0.4)	0.2 (<0.1)	0.2 (0.1)
VII	100 (0.3)	67 (13)	9 (0.3)	0.5 (<0.1)	0.5 (<0.1)
total	100	86	100	91	91

^acooked at 90 °C for 161 min.

^bcourgettes were blended at 80 °C for 15 min, mixed with uncooked ingredients (e.g. starch, milk proteins) and partitioned into 400 g packs. Note that units are batches before partitioning and packages after partitioning.

^cpasteurised at 90 °C for 4 min.

^dstored at 4 °C for 4 days in the factory, and then at 4 °C at retail, and 6.6 °C at home. Sell by date is 21 days.

^eB. cereus groups II, V and VI contain psychrotrophic strains.

4. Exposure assessment

The B. cereus hazards associated with minimally processed chilled foods correspond with two distinct exposures; exposure to spores (or less likely to cells) of psychrotrophic B. cereus and exposure to preformed cereulide toxin. Spores of B. cereus are present in the environment and therefore may enter the food chain. Spore germination in chilled foods, followed by cell multiplication and sporulation may lead to an increase in the spore concentration. Small spore exposures are likely to be frequent, but large exposures are expected to be rare. Consumption of an infectious dose of spores may ultimately lead to enterotoxin formation in the small intestine and diarrhoeal illness. Exposure to vegetative cells is less likely to lead to diarrhoeal illness. The ability to form cereulide toxin at ≤8 °C seems to be limited to a fraction of psychrotrophic B. cereus strains, and the quantity of cereulide formed has been low (Altayar & Sutherland, 2006; Guérin, Thorsen Rønning et al., 2017; Hoton et al., 2009; Thorsen et al., 2006). The thermal processes given to minimally processed chilled foods will not inactivate cereulide, thus when food is consumed, any cereulide toxin will pass into the GI tract and potentially cause emetic illness. A majority of commercially-produced minimally processed chilled foods are given heat treatments at 70 °C–90 °C, followed by a defined period of chilled storage typically at ≤8 °C (Peck et al., 2008). These foods have a strong safety record regarding B. cereus, with no reported outbreaks of foodborne illness when correctly stored (although the potential for under-reporting should be recognised). The following sections characterise the risk, identify the control measures that have contributed to prevent B. cereus causing diarrhoeal and emetic foodborne illness with correctly stored minimally processed chilled foods, and identify the critical factors associated with it.

5. Risk characterisation - psychrotrophic Bacillus cereus, correctly stored minimally processed chilled foods, and diarrhoeal illness

Diarrhoeal food poisoning is unlikely to be caused by ingestion of pre-formed enterotoxin in food. Although psychrotrophic B. cereus strains can produce diarrhoeal enterotoxins at refrigeration temperatures (Stenfors Arnesen, Granum, Buisson, Bohlin, & Nielsen-LeRoux, 2011; van Netten, van de Moosdijk, van Hoensel, Mossel, & Perales, 1990), these enterotoxins are very heat labile (inactivated by heating at 56 °C/5 min), and are therefore destroyed by thorough cooking prior to consumption (Ceuppens, Boon, & Uyttendaele, 2013; Ceuppens et al., 2011), and are also sensitive to acidity and proteolytic enzymes encountered in the stomach (Ceuppens, Rajkovic, et al., 2012; Ceuppens et al., 2011). EFSA considered pre-formed enterotoxins in food to be of little importance when assessing the risk of diarrhoeal infection, as the enterotoxins would be inactivated by the acidity and digestive enzymes in the stomach (EFSA, 2005).

For minimally processed chilled foods to be associated with diarrhoeal illness, psychrotrophic B. cereus needs to reach an infectious dose in the food and, once ingested, multiply and form enterotoxin in the small intestine at 37 °C. The lack of reported cases of diarrhoeal food poisoning associated with commercially-produced minimally processed chilled foods stored at the correct temperature is presumably because strains of psychrotrophic B. cereus have failed to reach a sufficiently high concentration to constitute an infectious dose (or have very rarely reached such a concentration, if there is under-reporting of illness). The total infective dose has been reported to be 10⁵-10⁸ cfu (EFSA, 2016; Stenfors Arnesen et al., 2008). Foods implicated in 84% of the diarrhoeal foodborne outbreaks investigated by Glasset et al. (2016) attributed to enterotoxigenic B. cereus strains had loads lower than 10⁵ cfu g⁻¹; and concentrations as low as 10² cfu g⁻¹ were recovered from some implicated foods. Given the need to survive passage through the stomach (see below), it is likely that the infectious dose for spores is less than that for vegetative cells.

From the data presented above, it is expected that the psychrotrophic B. cereus spore load of minimally processed chilled foods will be low, with the potential for spore germination and cell multiplication to an infectious dose during subsequent chilled storage. Since the infectious dose for spores is likely to be lower than that for vegetative cells (see below), a key issue is whether sporulation follows cell multiplication during chilled storage. Limited evidence indicates that sporulation of psychrotrophic B. cereus strains at chilled temperature is weak. For example, for two strains the minimum temperature for optimal sporulation (>90% phase-bright spores) was 15 °C (Gounina-Allouane, Broussolle, & Carlin, 2008), and B. weihenstephanensis strain KBAB4 sporulated with approximately 99% efficiency at 12°, 20° and 30 °C, but only 15% efficiency at 7° and 10 °C (Garcia, van der Voort, & Abee, 2010; Rajkovic, Kljajic, Smigic, Devlieghere, & Uyttendaele, 2013). Furthermore, when cells of three psychrotrophic enterotoxin-producing B. cereus strains were inoculated into spaghetti bolognaise and incubated at 12 °C, only one strain formed spores - 10⁴ spores g⁻¹ after 14 days incubation (Rajkovic et al., 2013). The sporulation capacity of B. cereus group strains is also reduced at the low oxygen concentrations likely to be encountered in minimally processed chilled foods (Abbas et al., 2014).

To cause illness, organisms need to survive a heat treatment applied to food immediately prior to consumption. The viability of B. cereus vegetative cells, but not spores (except, perhaps, for spores of some group VI strains), will be substantially reduced by a thorough heat treatment (Guérin, Dargaignaratz, Clavel, Broussolle, & Nguyen-the, 2017). Once ingested B. cereus has to survive passage through the GI tract where a variety of environmental insults are encountered, such as acidity in the stomach, exposure to bile salts and digestive enzymes, and competition with the host indigenous gut microbiome (Berthold-Pluta, Pluta, & Garbowska, 2015; Ceuppens, Uyttendaele, Hamelink, Boon, & Van de Wiele, 2012; Ceuppens, Van de Wiele et al., 2012). This passage is likely to significantly reduce the number of B. cereus vegetative cells entering the small intestine (Ceuppens, Uyttendaele, Drieskens, Rajkovic, et al., 2012; Ceuppens, Uyttendaele, Hamelink, et al., 2012; Ceuppens, Van de Wiele et al., 2012; Wijnands, Pielaat, Dufrenne, Zwietering, & Van Leusden, 2009), and may be one of the reasons why cell concentrations as high as 10⁸ cells ml⁻¹ failed to cause illness in a human volunteer study (Langeveld, van Spronsen, van Beresteijn, & Notermans, 1996). However, several studies have shown that B. cereus spores were not inactivated in simulated GI tract conditions (including ileum model), and could germinate and multiply to appreciable numbers in conditions simulating the ileum, depending on the extent of competition from the gut microbiome (Ceuppens, Rajkovic, et al., 2012; Ceuppens, Uyttendaele, Drieskens, Heyndrickx, et al., 2012; Ceuppens, Uyttendaele, Hamelink, et al., 2012; Ceuppens, Van de Wiele et al., 2012; Wijnands, Dufrenne, Zwietering, & van Leusden, 2006).

Given the short incubation period for the diarrhoeal syndrome (Table 1), strains that adapt rapidly to conditions in the GI tract and produce high quantities of enterotoxin will have the greatest potential to cause illness (Wijnands, Dufrenne, Zwietering, et al., 2006). The likelihood of a specific B. cereus strain causing diarrhoeal illness will reflect strain properties such as the hydrophobicity of the ingested spores (which determines attachment to epithelial cells), and the number and type of enterotoxin genes possessed and quantity of toxins formed (Andersson, Granum, & Rönner, 1998; Clavel, Carlin, Lairon, Nguyen-The, & Schmitt, 2004; Wijnands, Dufrenne, van Leusden, & Abee, 2007). Little is known about the ability of psychrotrophic B. cereus strains to colonise epithelial cells in the small intestine or what localised numbers are necessary to ensure sufficient enterotoxin production to cause diarrhoeal food poisoning. It is likely that B. cereus will not multiply in the lumen of the small intestine. Only B. cereus spores that attach to epithelial cells in the small intestine (or adhere to the mucus layer near epithelial cells) will germinate, and subsequently multiply and produce sufficient enterotoxin to cause diarrhoeal food poisoning (Berthold-Pluta et al., 2015; Ceuppens et al., 2013). Host-related factors, such as rates of gastric acid secretion and gastric emptying, bile and digestive enzyme secretion, composition of the gut microbiome, and nature of the food ingested will also influence survival and growth of ingested B. cereus (Ceuppens, Uyttendaele, Drieskens, Rajkovic, et al., 2012; Clavel et al., 2007; Da Riol, Dietrich, Märtlbauer, & Jessberger, 2018; Jeβberger et al., 2017).

A limited number of studies have investigated the ability of B. cereus strains to grow at both chilled temperature and 37 °C; approximately 54% (78 out of 145) of the strains examined in four studies could grow at both ≤8 °C and ≥37 °C (Dréan, McAuley, Moore, Fegan, & Fox, 2015; Miller et al., 2018; Stenfors Arnesen et al., 2011; Stenfors Arnesen, O'Sullivan, & Granum, 2007). However, the psychrotrophic B. cereus strains examined by Wijnands, Dufrenne, Zwietering, et al. (2006) that were able to grow at 37 °C, multiplied more slowly at this temperature than did mesophilic strains. There is also evidence that psychrotrophic strains have a lower potential to cause diarrhoeal illness. For example, only 27% (21 out of 78) of strains that grew at both ≤8 °C and ≥37 °C formed detectable enterotoxin or were cytotoxic in tests conducted at 37 °C (Dréan et al., 2015; Miller et al., 2018; Stenfors Arnesen et al., 2007, 2011), even though some strains demonstrated virulence at a lower temperature (e.g. 30 °C–32 °C). Recently, Miller et al. (2018) demonstrated that none of the 13 phylogenetic group VI strains tested at 37 °C were cytotoxic, even though all contained at least one enterotoxin gene. The nature of consumed food may also impact on enterotoxin formation and cytotoxicity (Da Riol et al., 2018).

In summary, for diarrhoeal food poisoning to be associated with correctly stored commercially-produced minimally processed chilled foods, psychrotrophic strains of B. cereus need to survive processing hurdles and multiply in chilled food to a concentration capable of causing infection after ingestion. Strains of only three B. cereus phylogenetic groups (II, V and VI) have psychrotrophic properties, defined here as the ability to multiply at ≤8 °C. An EFSA opinion on B. cereus and various publications have questioned the pathogenicity of psychrotrophic B. cereus strains that grow at ≤8 °C and have indicated uncertainty concerning their ability to cause diarrhoeal illness and also what constitutes an infectious dose (Choma et al., 2000; EFSA, 2005; Guinebretière et al., 2008; Miller et al., 2018; Stenfors et al., 2002). These reports indicate that psychrotrophic B. cereus strains may not be adapted to grow in the small intestine and produce insufficient enterotoxin to cause illness. The infectious dose for at least some strains of psychrotrophic B. cereus is therefore likely to be greater than that for mesophilic strains.

Differences can be identified in the hazard presented by the three phylogenetic groups of psychrotrophic B. cereus. In particular, Group VI strains have a lower spore heat resistance, grow better at chilled temperatures, and are less pathogenic (probably with a higher infectious dose (Afchain et al., 2008; Carlin et al., 2013; Guinebretière et al., 2010; Luu-Thi et al., 2014) than strains of groups II and V. Thus, the heat process applied to the food, and the period and temperature of chilled storage are likely to affect the relative risk presented by the three phylogenetic groups of psychrotrophic B. cereus. Overall it appears that group II strains present the greatest risk, followed by group V strains, with group VI strains representing the lowest risk of diarrhoeal food poisoning (Afchain et al., 2008; Carlin et al., 2013; Guinebretière et al., 2008, 2010; Luu-Thi et al., 2014).

The lack of association between diarrhoeal food poisoning and correctly stored commercially-produced minimally processed chilled foods indicates that an infectious dose has not been reached, and assuming that diarrhoeal food poisoning is fully reported and given that more than 10¹⁰ packs of chilled food have been sold in the UK alone, then the risk presented is very low. However, further risk characterisation requires additional quantitative data. For example: (i) loading of B. cereus groups II, V and VI in raw materials used in minimally processed chilled foods; (ii) growth of B. cereus groups II, V and VI at chilled temperatures, (iii) determination of an infectious spore dose; (iv) sporulation of B. cereus groups II, V and VI in minimally processed chilled foods, and (v) ability of cells of B. cereus groups II, V and VI to survive GI tract passage, colonise the intestinal epithelia and form enterotoxins in vivo (note that if ingested B. cereus vegetative cells are unable to bring about diarrhoeal illness, and if there is no sporulation in food, then the hazard is independent of product storage time).

6. Risk characterisation - psychrotrophic Bacillus cereus, correctly stored minimally processed chilled foods, and emetic illness

Emetic food poisoning is caused by ingestion of foods containing pre-formed cereulide; an extremely heat stable cyclic peptide toxin that is highly resistant to acid, alkali and proteolytic enzymes (Table 1). Cereulide formation by B. cereus in food poses a hazard to health as the pre-formed toxin will survive both heat treatments given to the food and passage through the GI tract (Ceuppens et al., 2011). The toxic dose of cereulide for humans is estimated to be 8 μg kg⁻¹ body weight (Jääskeläinen et al., 2003), thus a 50 kg human will need to ingest approximately 400 μg of cereulide to cause illness. In recent outbreaks of emetic food poisoning associated with foods exposed to ambient temperatures, it has been estimated that the foods contained 0.01–85 μg cereulide g⁻¹ of food (Agata et al., 2002; Naranjo et al., 2011; Yamaguchi, Kawai, Kitagawa, & Kumeda, 2013). The lower estimates do not seem consistent with the estimation of the human toxic dose.

The absence of reported cases of emetic food poisoning associated with correctly stored minimally processed chilled foods is presumably because strains of psychrotrophic B. cereus have failed to form sufficient cereulide in these foods to bring about an intoxication. The ability to produce cereulide has been described in a small number of psychrotrophic B. cereus group VI strains (Altayar & Sutherland, 2006; Carroll, Kovac, Miller, & Wiedmann, 2017; Hoton et al., 2009; Thorsen et al., 2006):

• •Thorsen et al. (2006) quantified cereulide formation by two strains of B. weihenstephanensis (MC67 and MC118) following incubation on BHI agar for 10 days at temperatures from 8° to 25 °C. For each strain, more than a thousand-fold more cereulide was formed at 25 °C than at 8 °C. These strains produced 1.4–2.1 μg cereulide g biomass⁻¹ in 10 days at 15 °C, and 0.1 μg cereulide g biomass⁻¹ in 10 days at 8 °C (Thorsen et al., 2006). Assuming that 1 g of biomass equals 10¹⁰ cfu (Yu, Szymanowski, Myneni, & Fein, 2014), and the viable count reached 10⁸ cfu g⁻¹ in a food, then the maximum cereulide concentration would be 0.02 μg g⁻¹ at 15 °C, and 0.001 μg g⁻¹ at 8 °C.
• •Cereulide formed by B. weihenstephanensis strain MC118 in a cooked meat sausage after 4 weeks incubation at 8 °C was below the limit of quantification (1 ng g⁻¹ meat) (Thorsen, Budde, Koch, et al., 2009).
• •B. weihenstephanensis strain MC67 formed cereulide at 25 °C during late exponential growth phase and continued into stationary phase during surface growth on agar plates (Thorsen, Budde, Henrichsen, Martinussen, & Jakobsen, 2009). However, after incubation for 1 week at 8 °C, cereulide production was low (2 ng cereulide cm⁻² on an agar plate) at a cell density of ≈10⁸ cfu cm⁻². No further increase in cereulide concentration was detected after an additional 3 weeks incubation at 8 °C. Assuming 1 cm² agar surface is comparable to 1 g of food (Thorsen, Budde, Henrichsen, et al., 2009), then a final concentration of 0.002 μg g⁻¹ was reached.
• •Guérin, Thorsen Rønning et al. (2017) analysed cereulide formation by B. weihenstephanensis strains BtB2-4 and MC67 during surface growth on agar plates. Growth was removed and cereulide formation quantified. Assuming growth covered the entire petri dish, and that 1 cm² agar surface is comparable to 1 g of food, then strain BtB2-4 produced quantifiable amounts of cereulide (0.017 ng g⁻¹ food) after 7 days incubation at 8 °C, when growth had reached early stationary phase. Strain MC67 reached stationary phase later (9–10 days at 8 °C), and only produced detectable quantities of cereulide (0.001 ng g⁻¹ food) at day 11 (Guérin, Thorsen Rønning et al., 2017).

It should be noted that since cereulide formation by both mesophilic and psychrotrophic B. cereus group strains occur during late exponential growth phase and continues during stationary phase (Biesta-Peters et al., 2016; Ceuppens et al., 2011; Thorsen, Budde, Henrichsen, et al., 2009), then a longer period of incubation may have led to an increase in cereulide formation. However, the limited data presently available indicate that only a low concentration of cereulide may be produced at 8 °C by strains of psychrotrophic B. cereus, and although this present information has limitations, it may explain why correctly stored minimally processed chilled foods have not been associated with emetic food poisoning.

Differences can be identified in the hazard presented by the three phylogenetic groups of psychrotrophic B. cereus, as with diarrhoeal food poisoning. Group VI strains have a lower spore heat resistance but grow better at chilled temperatures, than strains of groups II and V. (Afchain et al., 2008; Carlin et al., 2013; Guinebretière et al., 2010; Luu-Thi et al., 2014). Thus, the heat process and the period/temperature of chilled storage are likely to affect relative risk presented by the three groups. An additional important factor is the ability of strains of the three phylogenetic groups to form significant quantities of cereulide at chilled temperatures. To date, all psychrotrophic B. cereus strains that form emetic toxin belong to group VI, and thus may pose a greater risk than strains from groups II and V. Additional quantitative data on cereulide formation will support this assessment.

In summary, cereulide will survive both the heat treatment applied to foods and passage through the GI tract. The risk of emetic food poisoning is more readily characterised than that of diarrhoeal food poisoning, as a greater amount of quantitative data are available. The limited ability of psychrotrophic B. cereus strains to form cereulide at 8 °C and below, and the subsequent small quantity formed is the most likely explanation for the lack of reported cases of emetic food poisoning associated with correctly stored commercially-produced minimally processed chilled foods. However, it should be noted that existing strain collections may not represent the entire range of environmental isolates, and the possibility remains that presently undescribed (or untested) strains of psychrotrophic B. cereus may form significant quantities of cereulide at ≤8 °C, and there also exists the potential for psychrotrophic B. cereus group strains to acquire the ces genes by horizontal gene transfer (Mei et al., 2014).

7. Conclusions

There have been no reported cases of diarrhoeal or emetic food poisoning associated with psychrotrophic B. cereus and correctly stored commercially-produced minimally processed chilled foods. In the UK alone more than 10¹⁰ packs of these foods have been sold. Although these types of food poisoning may not be fully reported, the risks from B. cereus and correctly stored commercially-produced minimally processed chilled foods appear small. Psychrotrophic B. cereus strains may be more likely to be a food spoilage risk rather than food poisoning risk with minimally processed chilled foods stored correctly.

The absence of reported diarrhoeal illness associated with correctly stored commercially-produced minimally processed chilled foods indicates that strains of psychrotrophic B. cereus have never (or very rarely) reached an infectious dose. This could be because: (i) only a fraction of B. cereus populations (specifically groups II, V, and VI) are both pathogenic and able to grow at refrigeration temperatures (≤8 °C); and (ii) only a fraction of B. cereus populations can survive food processing, multiply in the minimally processed chilled food, survive passage through the GI tract, and then go on to produce sufficient toxin in vivo in a relatively short time. However, quantification of the risk is challenging because there are limited data concerning the prevalence of the different B. cereus phylogenetic groups in foods and the statistical dependency associated with growth at refrigeration temperature is unclear. Additionally, there is substantial uncertainty associated with; (i) what constitutes an infectious dose for psychrotrophic B. cereus strains, and (ii) what is the ability of psychrotrophic B. cereus to survive gastric passage and colonise the intestinal epithelia and then form enterotoxins in vivo.

The absence of reported emetic illness associated with correctly stored commercially-produced minimally processed chilled foods indicates that strains of psychrotrophic B. cereus have never (or very rarely) formed a toxic dose of cereulide. This is likely to be because at chilled temperatures (8 °C and below) very few psychrotrophic B. cereus strains form cereulide, and for those that do the quantity formed is low, making the development of a toxic dose in food very unlikely. This is consistent with the observed pattern of illness associated with emetic food poisoning, where failure to cool food sufficiently or store food at a sufficiently high temperature are often identified as factors in causing emetic intoxication.

The strong safety record of correctly stored commercially-produced minimally processed chilled foods in relation to B. cereus food poisoning may be related to a combination of low pathogenicity of psychrotrophic B. cereus at chilled temperature, good quality raw materials with low B. cereus spore loading, high manufacturing hygiene standards, chilled storage conditions maintained throughout the supply and retail chain and a restricted shelf-life.

Appendix A. Supplementary data

The following is the Supplementary data to this article: