Tracking emerging pathogens: the case of noroviruses

4.1. Introduction

Viruses are increasingly recognised as pathogens involved in foodborne infections. One of the reasons for this increased awareness is the improved laboratory capability to detect the groups of viruses causing gastroenteritis, the most common foodborne illness worldwide. Only ten years ago, over 90% of community cases and outbreaks of gastroenteritis went without diagnosis in The Netherlands. Within a decade this proportion of unknowns was reduced to 54% and less than 20% for population or outbreak studies respectively (de Wit et al., 2001a; van Duynhoven et al., 2005). These improvements resulted largely from the recognition of noroviruses (NoVs, formerly known as Norwalk like viruses or small round structured viruses) as the single most important cause of gastroenteritis both in community cases and outbreaks. Data from epidemiological studies, outbreak reports, and (international) surveillance are building the case for the role of food-and waterborne transmission in the epidemiology of these viruses.

The increased national and international surveillance activities have shown that viruses are not only more often detected than in the past but have also recognised that several new virus strains have been truly emerging. International outbreaks of gastrointestinal illness have been caused by NoV strains, which have evolved by recombination or genetic drift. While other viruses, such as hepatitis A and hepatitis E viruses, astroviruses, rotaviruses, enteric adenoviruses, and some enteroviruses (Seymour and Appleton, 2001; Koopmans and Duizer, 2004) can be foodborne, this chapter will deal mainly with the NoVs. It should be noted, however, that what we have learned from these nowadays easily recognised pathogens, may be true for several other structurally similar viruses, some of which cause more serious diseases. Some of these concerns will be discussed in Section 4.7.

4.1.1. The noroviruses

Viruses are unique in nature. They are the smallest of all self-replicating organisms, historically characterised by their ability to pass through filters that retain even the smallest bacteria. In their most basic form, viruses consist solely of a small segment of nucleic acid encased in a simple protein shell. They have no metabolism of their own but rather are obliged to invade cells and parasitize subcellular machinery, subverting it to their own purposes. Many have argued that viruses are not even living, although to a seasoned virologist they exhibit a life as robust as any other creature. (Condit, R. C. Principles of Virology, in Fields Virology (2001), with permission.)

This definition naturally covers the NoV and is a good starting point for consideration of viruses in matters of food safety. NoV are human enteric caliciviruses. The family Caliciviridae belongs to the picornavirus-like superfamily; small, RNA viruses without an envelope. The calicivirus particles (virions) measure between 27 and 38 nm and consist of a spherical protein shell and a genome of a single strand of approximately 7.6 kb positive sense RNA. The caliciviruses known to infect humans belong to two genera: the genus Norovirus and the genus Sapovirus (SaV, formerly known as Sapporo Like Viruses, or typical caliciviruses). In addition, the family holds two genera of animal viruses, named Vesivirus and Lagovirus (Green et al., 2001). A recently detected bovine calicivirus may be assigned to a fifth genus (Smiley et al., 2003). The genus NoV is further subdivided into an increasing number of poorly defined genogroups (Karst et al., 2003). Each genogroup can be subdivided in several genotypes, defined by > 80% amino acid homology across the capsid gene, and over 15 genotypes are characterised at the moment (Koopmans et al., 2003). The NoV are officially denoted by the following cryptogram: host species/genus abbreviation/species abbreviation (i.e. genogroup)/strain designation/year of detection/country of origin. For NoVs the strain designation is the location where the strain was first detected, for example: Hu/NV/GGI/Norwalk Virus/1968/US, Hu/NV/GGI/Southampton virus/1991/UK, Hu/NV/GGII/Hawaii virus/1971/US and Hu/NV/GGII/Bristol/1993/UK (Green et al., 2001).

Even though in vitro culture systems are not yet available for the NoV (Duizer et al., 2004b), a considerable quantity of data has been obtained on the structure of the genome and the capsid of these viruses. The NoV genome is organised into a 5′untranslated region (UTR), the open reading frames (ORF) 1, 2, and 3, a 3′UTR and a poly A tail. ORF1 encodes the non-structural proteins necessary for virus replication. The major structural protein (VP1) is coded on ORF2, whereas ORF 3 encodes a minor structural protein (VP2) (Glass et al., 2000; Bertolotti-Ciarlet et al., 2003). Altogether the virion consists of a capsid of 180 copies of VP1 and a small number of VP2, and the single stranded RNA genome which is possibly covalently linked to the capsid by a VPg protein (Green et al., 2001).

Being such small and ‘simple’ organisms, with a small genome, viruses rely to a great extent on the enzymes in the host cell they invade to undergo their replicative cycle. Unlike bacteria, viruses are obligate intracellular parasites. Many viruses, including the human enteric caliciviruses, have a narrow species- and tissue tropism, i.e., efficient replication of the human NoVs occurs only in the gastrointestinal tract of humans. Replication in other human tissues has not been detected, and even though NoV genotypes were found in cattle (e.g. bovine Jenavirus, GGI) and pigs (e.g. porcine enteric calicivirus, GGII), none of these animal genotypes have ever been detected in human patients.

4.1.2. Symptoms of a calicivirus infection

The NoV are the major causative agents of classical viral gastroenteritis in all age groups (De Wit et al., 2001a). After a short incubation period of 12-48 hours, the disease is characterised by an acute onset of symptoms as (non-bloody) diarrhoea, vomiting, (low-grade) fever, nausea, chills, weakness, myalgia, headache and abdominal pain. Hence, the illness is referred to as ‘gastric or stomach flu’. The attack rate of NoVs is typically around 45% or higher, but this differs with virus genotype and is also dependent on the host's genetic susceptibility. Most notable is the usually sudden onset of projectile vomiting, resulting in efficient spreading of viruses. Although patients do, in general, feel acutely ill, and dehydration can be severe, the illness is rarely fatal and subsides on average after two to six days, with the longest duration of illness in children (Rockx et al., 2002). A typical NoV infection is self-limiting and treatment focuses on supportive care and prevention and treatment of dehydration.

No antiviral treatment has been found to be effective in treating NoV infection and no vaccine is available yet, although clinical trials with recombinant vaccines based on the capsid protein of NoV are in progress. For the NoV and SaV no long-term sequellae have been reported, but the duration of the infectious period might have been underestimated. Recently it was found that shedding of NoV lasted at least a week, but continued for three weeks in 26% of the patients (Rockx et al., 2002). In immuno-compromised persons, chronic infection may develop with persistent diarrhoea and long-term shedding (Nilsson et al., 2003; Gallimore et al., 2004b).

Immune response and genetic susceptibility

Infected individuals do develop short-term immunity to homologous virus (up to 14 weeks), but the existence or development of long-term immunity to NoV is still elusive (Matsui and Greenberg, 2000). Pre-existing antibodies do not always correlate with protection from infection, and insight into the correlation between NoV genotypes and serotypes is only just beginning to develop. Serum IgG, IgA and IgM responses are observed after infection with NoV and common epitopes within genogroups and between genogroups have been identified (Erdman et al., 1989a and b; Gray et al., 1994; Treanor et al., 1993; Hale et al., 2000; Yoda et al., 2000; Harrington et al., 2002; Kitamoto et al., 2002). However, there are contradictory data on the level of cross-reactivity after NoV infections in humans. It has been shown that the specificity of the immune response varies greatly within genogroups but nevertheless, other studies have observed cross reactivity between viruses belonging to different GGs (Farkas et al., 2003; Madore et al., 1990; Treanor et al., 1993).

Recent studies show that sensitivity to NoV infection is dependent on ABH-histo-bloodgroup antigens (carbohydrates), Lewis antigens and secretor status (Hutson et al., 2002; Marionneau et al., 2002; Huang et al., 2003; Harrington et al., 2004; Rockx et al., 2005) and that the genetically determined susceptibility is different for different genotypes. For example, bloodgroup B, and non-secretor status confer protection to infection by Norwalk virus, bloodgroup A confers protection to infection by Snow Mountain virus, while none of the bloodgroups offers protection to infection with GGII.4 viruses (reviewed by Hutson et al., 2004). These studies also suggest that the susceptibility to NoV infection is related to specific carbohydration of the host receptor, and that NoV binding can be inhibited by the specific carbohydrates in solution. These findings may lead to the development of (strain specific) antiviral treatments. Research into the genetic susceptibility is booming at the moment and insight may develop and change over time.

4.1.3. Epidemiology of viral gastroenteritis and examples of viral foodborne outbreaks

The NoVs are transmitted by the faecal-oral route; they are shed in vomitus and faeces and enter their next victim orally. In persons with clinical illness, viruses typically are detected in stools at levels far exceeding 10⁶ virus particles/ml (the titers for vomit are not known). Since the minimal infectious dose of the NoVs is believed to be very low (between 1-10 particles), and immunity short-lived, introduction of NoV in a population easily leads to an outbreak, affecting many people. Even though the attack rate is rather high, asymptomatic infections are quite common. For example, in a community study in the Netherlands, evidence of virus shedding was found in 5.2% of controls, i.e., persons without gastrointestinal complaints (de Wit et al., 2001a) and in 19% of people without gastrointestinal illness in an outbreak setting (Vinje et al., 1997). The efficient spread of viruses by (projectile) vomiting, the prolonged shedding in faeces after recovery from illness, and the shedding of viruses from asymptomatic carriers are important factors contributing to the impressive numbers of NoV infections.

Burden of disease studies

Few studies have looked at the incidence and health impact of NoV infection at the community level. The most extensive data are from the UK (Tompkins et al., 1999; Wheeler et al., 1999) and The Netherlands, where a randomised sample of the community participated in cohort studies of infectious intestinal disease (IID). The incidence of community-acquired IID was calculated as 190 per 1000 person years in the UK and 283 per 1000 person years in The Netherlands (Tompkins et al., 1999; de Wit et al., 2001a). Viruses were the most frequently identified causes of community acquired gastroenteritis, with NoV detected in 11% of cases in The Netherlands and 7% in the UK. This difference may partly result from the different methods used for virus detection: de Wit et al. used RT-PCR whereas the study in the UK employed the far less sensitive electron microscopy.

In both studies, the referral of patients to a general practitioner was studied as well. Approximately 5% of cases in The Netherlands sought treatment, compared with 4% of cases in Wales and 17% in England. The physician-based patient group was younger, and had more severe symptoms and longer duration of illness (de Wit et al., 2001b). Some 5% of physician-based cases were NoV positive in The Netherlands, and 6.5% in the UK. The lower proportion of NoV disease in this study population compared with the community cases confirms that on average NoV illness is relatively mild. The financial consequences of this mild disease can, however, be serious; cost estimates related to NoV infections for the Netherlands showed that 13% (i.e. 46 million) of costs for gastroenteritis were due to NoV infections alone (van den Brandhof et al., 2004).

Smaller studies in selected patient populations have been conducted elsewhere, and show that NoV are known to occur as a prominent cause of illness in countries throughout Europe, the USA, Australia, Hong Kong and Japan (Fankhauser et al., 1998, 2002; Lopman et al., 2002, 2003, 2004; Marshall et al., 2003; Lau et al., 2004; Iritani et al., 2002, 2003). Additionally, evidence is mounting that the disease may be common in countries with different degrees of development across the world, with studies from, for example, Hungary, Argentina, Brazil, Pakistan and India (Farkas et al., 2002; Reuter et al., 2003; Martinez et al., 2002; Parks et al., 1999; Gallimore et al., 2004a; Phan et al., 2004; Girish et al., 2002). NoV infection is common in all age groups but the incidence is highest in young children (< 5 yrs).

Outbreak studies

Probably the best known presentation of NoV is that of large outbreaks of vomiting and diarrhoea that lend the disease the initial description of ‘winter vomiting disease’ Zahorsky, 1929; Mounts et al., 2000). Since the development of molecular detection methods NoV have emerged as the most important cause of outbreaks of gastroenteritis in institutional settings (i.e. hospitals, nursing homes). The majority of NoV gastroenteritis cases results from direct person-to-person transmission. However, NoV-related outbreaks have been shown to be food- or waterborne caused by, for example, contaminated shellfish (Doyle et al., 2004; Kingsley et al., 2002a; Le Guyader et al., 2003), raspberries (Ponka et al., 1999) or drinking water (Carrique-Mas et al., 2003; Kukkula et al., 1999; Parshionikar et al., 2003). Additionally, environmental spread of NoV was found, for instance by contaminated carpets in hotels (Cheesbrough et al., 2000), toilet seats and door handles in a rehabilitation centre (Kuusi et al., 2002), and contaminated fomites on hard surfaces, carpets and soft furnishings in a concert hall (Evans et al., 2002).

In The Netherlands, approximately 12-15% of community cases of NoV gastroenteritis was attributed to foodborne transmission, based on analysis of questionnaire data. This makes NoV as common a cause of foodborne gastroenteritis as Campylobacter, and more common than Salmonella (de Wit et al., 2003).

Outbreaks are quite often a result of a combination of several transmission routes, for example introduction of the virus in a sensitive population by food, water or an asymptomatic shedder, followed by efficient spread of the virus through the susceptible population by direct person-to-person transmission. Large food- or waterborne outbreaks due to a common (point) source introduction are less common than multiple transmission routes outbreaks, but they do occur (see, for example, Cannon et al., 1991; Kohn et al., 1995; Daniels et al., 2000; Girish et al., 2002). Several waterborne outbreaks have been reported as a result of contaminated private wells or communal water systems in Sweden (Carrique-Mas et al., 2003; Nygard et al., 2003), municipal water in Finland (Kukkula et al., 1999), and well water in Wyoming USA (Anderson et al., 2003). Interestingly, no signs of faecal contamination by testing for indicator bacteria were found by Carrique-Mas and co-workers (Carrique-Mas et al., 2003).

Foods can be contaminated with NoV anywhere along the food chain from farm to fork. Wherever a NoV carrier comes in contact with food, contamination might occur and due to stability of these pathogens, they are likely to survive many food processes (see Section 4.6; Koopmans and Duizer, 2004). When viral contamination occurs through a person touching food, the contamination will be localised in spots (focally). Infections caused by focally contaminated foodstuffs are most likely to be recognised as foodborne when the contamination has occurred at the end of the food chain. Two reported examples show the characteristics of such outbreaks. One report from Sweden describes the large-scale outbreak detected in 30-day care centres in the Stockholm area in March 1999. All centres obtained their lunch meals from one caterer, and approximately 25% of all customers (n = 1500 customers) fell ill. The early cases had an average incubation time of 34 hours but the outbreak went on for at least 12 days, due to secondary infections. One of the food handlers fell ill in the same time window as the early cases and none of the other food handlers reported symptoms. This implies that the most likely route of food contamination was by a pre- or asymptomatic food handler (Gotz, 2002).

Another large point-source foodborne outbreak occurred in the Netherlands in January 2001. A baker who had been sick with vomiting the previous day had prepared the rolls for the buffet lunch at a New Year's reception. Within 50 hours after the reception, over 200 people had reported ill with gastrointestinal symptoms. Epidemiological and microbiological investigations indicated the rolls as point source of this outbreak with a total of 231 people sick with diarrhoea and vomiting (de Wit et al., in press). It was noted that the baker was still ill during preparation of the rolls and that he vomited in the sink. Cleaning the sink with chlorine did not prevent the outbreak. It is probable that the spread of infectious viruses by vomiting was not restricted to the sink or the cleaning method used was not stringent enough to decontaminate the sink of these resistant viruses.

In many outbreaks, however, the seeding event is not recognised. This can be on two levels; (i) the route of introduction of the NoV to the affected population is unknown, or (ii) the vehicle of introduction is known but it is unclear how and when that vehicle (vector) was contaminated. Resolving the transmission routes and vectors involved in such outbreaks is the challenge for viral food safety, now and in the near future.

4.2. Detection

To date, laboratory diagnosis of NoV is based on molecular biological techniques and electron microscopy (EM). Both techniques are based on the detection of the pathogen in stool samples, rather than on measuring an immune response of the host to pathogen. EM is a ‘catch all’ method, which allows detection of many viruses as well as basic typing of viruses. On the basis of immuno-EM, in which samples are treated with specific antisera, NoV antigenic types have been defined which correlate with the genotypes established following later genomic characterisation (Lewis, 1990; Green et al., 1995). EM is, however, quite insensitive (detection limit of 10⁵ to 10⁶ particles per gram stool), and requires specialised equipment and highly skilled personnel.

Presently, molecular biological techniques (RT-PCR) are used to detect the viral genome (nucleic acid, RNA in NoVs), and since they can be very sensitive and allow for typing to strain level, this is the method of choice. In addition, some antigen detection based methods (ELISA, EIA) are gaining popularity. These assays have the advantage of simplicity, and many samples can be screened fast. However, the currently available EIAs are less sensitive than RT-PCR methods (Rabenau et al., 2003; Richards et al., 2003; Burton- Macleod et al., 2004). Unfortunately, no culture method is available, significantly hampering studies of the infectivity of the detected viruses.

4.2.1. Viral gastroenteritis: the Kaplan criteria

Since the discovery of Norwalk virus (the prototype strain of the NoV) by Kapikian in 1972 using immune electron microscopy (Kapikian et al., 1972), the role of these viruses in acute gastroenteritis was becoming more clear every year. In a thorough investigation of the role of Norwalk virus in outbreaks of acute nonbacterial gastroenteritis in 1982, Kaplan and co-workers distilled a set of common features of viral gastroenteritis outbreaks. This set of criteria is now known as the Kaplan criteria and may be used as an epidemiological tool to diagnose outbreaks as caused by NoV (Kaplan et al., 1982; Hedberg and Osterholm, 1993; Lopman et al., 2002; Koopmans et al., 2002).

The Kaplan criteria used to ascribe gastroenteritis outbreaks to a NoV infection are as follows:

• •stools negative for bacterial and parasitic pathogens
• •proportion of cases with vomiting > 50%
• •mean duration of illness 12-60 hours
• •mean incubation period 24-48 hours
• •high attack rate and high number of secondary cases.

It is clear that this method does not allow for virus typing, detailed epidemiological studies or single patient diagnostics. However, for countries without laboratory capacity for NoV diagnosis, the use of Kaplan criteria may give a first indication of the proportion of outbreaks that are likely to be viral.

4.2.2. Detection and typing in patients: RT-PCR and ELISA

Even though there is no real treatment for noroviral gastroenteritis, quick laboratory diagnosis can help control the spread of the disease and prevent the ineffective use of antibiotics. Currently the method of choice for the diagnosis of NoV infection is the molecular biological technique called reverse transcription polymerase chain reaction (RT-PCR). This method is based on the detection of viral nucleic acid in faecal samples. The high sensitivity of this method (detection limit of 10-100 viral particles per gram stool) is one of its major advantages. Another advantage is that – theoretically – it can be applied to all kinds of substrates such as faeces, vomitus, serum, food-matrices and water. A RT-PCR protocol involves, in short, three stages:

• 1.RNA extraction from the matrix: the viral RNA has to be released from the capsid and the resulting RNA suspension should be cleared from RT-PCR inhibitors, (RNases and DNases). The RNA extraction step forms the most crucial difference between methods for virus detection in food, water or environmental samples and clinical specimens. The importance of this step will be discussed in greater detail in section 4.2.3.
• 2.The RT and PCR steps, in which a fragment of the viral RNA genome is transcribed into DNA and then amplified.
• 3.Confirmation of the RT-PCR product to rule out false positive results and to type the amplified fragment.

The optimisation of stages 2 and 3 of this method have been hampered by the high genetic diversity of the NoV. At present, several protocols are used which have been developed for the detection of a broad range of virus strains but still, no single primer pair can detect all NoV strains (reviewed by Atmar and Estes, 2001). Additionally, generic tests are optimised for the detection of a broad range of viruses, not for the specific and more sensitive detection of one strain (Vennema et al., 2002). While this may be appropriate for patient diagnostics (where high levels of virus are shed in the stool samples), virus detection in environmental samples, food or water requires highly optimised assays with low detection limits. This is not compatible with broad range detection.

The most frequently used diagnostic RT-PCR protocols amplify either a part of the gene for the RNA-dependent RNA polymerase (RdRp), located near the 5′ site of ORF1 of the NoV genome, or a region at the 5′- end of capsid protein encoding ORF2. Both parts were shown to be conserved enough for the development of primer sets able to detect a range of viruses (generic test), yet variable enough for strain typing (Vennema et al., 2002; Vinje et al., 2004). Others have suggested amplifying other parts of the genome, for example a region at the 3′- end of ORF1 (Fankhauser et al., 2002) or a region at the 3′- end of ORF2 (Vinje et al., 2004). However (as will be explained in detail in Section 4.3 of this chapter) for detection, typing and research on transmission routes, harmonisation of methods is urgently needed. Therefore a continuous changing to different genome regions for typing should not be encouraged.

The typing of NoV is very important for studies of the transmission routes, and, for example, to establish common source outbreaks (Koopmans et al., 2003). Several typing methods have been described (reviewed by Atmar and Estes, 2001) but the most relevant for virus tracking is the most elaborate; sequencing of the amplified genome fragment. Sequence data and phylogenetic analysis of the RdRp fragment and a fragment of the gene coding for the capsid protein (ORF2) have revealed that recombinant NoV strains have evolved, and thus that analysis of more than one region of the NoV genome may be important (Vinje et al., 2003; Atmar and Estes, 2001). An interesting method for typing of recombinants is described by Kageyama et al. (2003), who suggest amplification of the ORF1-ORF2 junction region, a fragment overlapping the RdRp and capsid gene. In that way, sequencing of just one fragment should suffice to type the strain and to discriminate between recombinants and ‘old’ strains. However, this will apply only when the recombination event has occurred in the junction region. There is at this time not enough information available to assume the junction region is the only recombination site.

Recent developments in molecular biology techniques have yielded several new diagnostic tests. One is the development of real-time RT-PCR methods, which are faster than the original RT-PCR protocols, mainly due to the fact that the RT-PCR and confirmation stages are incorporated in one step (Richards et al., 2004; Kageyama et al., 2003). Another new method is the nucleic acid sequence based amplification (NASBA) technique which, in theory, should be simpler to perform and at least as sensitive as the RT-PCR (Jean et al., 2003; Moore et al., 2004). However, validation of these newer tests is still incomplete, rendering the specificity, sensitivity and broadness largely unknown. Whichever genome detection assay is used for virus detection, one should realise that false negative tests are relatively common, due to the great genetic diversity resulting in primer mismatches (Vinje et al., 2003). Therefore, back-up protocols should be used for unexplained outbreaks that do meet the Kaplan criteria.

The NoV are not only genetically but also antigenically very diverse with as much as 30% variation in amino acid sequence of the major capsid proteins within the genogroups and up to 50% between genogroups. This great diversity has hampered the development of broadly reactive diagnostic tests for antigen detection. After it was discovered that expression of the NoV ORF2 by recombinant baculoviruses in insect cells resulted in the self-assembling of the capsid protein into virus like particles (VLPs, Green et al., 1993), many antigen-based detection methods have been developed. The VLPs were first used to detect antibodies to NoV in serum samples (Green et al., 1993; Parker et al., 1993). Later, sera from animals inoculated with VLPs were used to develop enzyme immunoassays (EIAs) for the detection of NoVs in faeces (Jiang et al., 1995; Herrmann et al., 1995). Currently, several EIAs to detect NoV antigens are commercially available (Richards et al., 2003; Burton-Macleod et al., 2004); they are slightly less sensitive than RT-PCR methods, but since the EIAs can be applied by non-specialised laboratories and results can be obtained within 3-4 hours, these assays will be increasingly used to diagnose gastroenteritis outbreaks. The EIAs do not, however, yield information on virus typing, other than possibly the discrimination between GGI and GGII. Therefore, RT-PCR and sequencing of positive outbreaks will still have to be done to gain the information needed for virus tracking and to pinpoint common-source outbreaks.

4.2.3. Detection in foods and water

Detection of NoVs in foods and water also relies mostly on RT-PCR methods. Due to the enormous variety in food matrices a considerable challenge for food microbiologists is found in the first stage of the RT-PCR protocol; the extraction of the viral RNA from the matrix. Many studies have focused on method development to extract viral RNA from shellfish, which has resulted in a variety of methods as reviewed by Lees (2000). Common features of these are dissection of the digestive tract and hepatopancreas (Schwab et al., 1998) and subsequent homogenisation, followed by variable (partial) purification and RNA extraction methods. The methods applied are however laborious and specialised and therefore, in general, not available to most routine labs. Other matrices for which virus detection methods have been developed include fresh produce and fruits (Bidawid et al., 2000; Leggitt and Jaykus, 2000; Dubois et al., 2002; Sair et al., 2002; Le Guyader et al., 2004), and hamburgers, ham, turkey and roast beef (Leggitt and Jaykus, 2000; Schwab et al., 2000; Sair et al., 2002). Since foods are often contaminated by an infected food handler, the contamination of food items may be low level and focal.

Contamination with irrigation water may result in more diffuse presence of viruses, but primarily at the surface of the produce or product. Therefore most protocols are based on elution of the virus particles from the surface of the product, followed by a concentration step (mostly ultra-centrifugation or -filtration) and RNA extraction from the concentrate (Le Guyader et al., 2004). Recently an immunomagnetic capture method was used to capture (concentrate and purify) NoV from a crude food suspension (Kobayashi et al., 2004). This method at present cannot be generalised because antibodies to NoV are not broadly reactive although some cross-reactivity exists between genotypes within the same genogroup. Further evaluation is needed to see if sufficiently broad reactivity exists.

For NoV detection in water a variety of methods is available (Gilgen et al., 1997; Straub et al., 2003; Karim et al., 2004; Haramoto et al., 2004). Here, too, the challenge exists in obtaining concentrates with a detectable level of viral RNA, but a low level of RT-PCR inhibitors. Many protocols are variations on a general method which, in short, entails concentration by adsorption to a filter (e.g. a positively charged membrane, glasswool), subsequent elution from the filter and further concentration using a microconcentrator or precipitation step. But immunoaffinity concentration and purification protocols have been developed for application in water too (Schwab et al., 1996).

The difficulties in food safety issues regarding the cumbersome detection of NoVs (or HAV) in shellfish and water is further complicated by the recognition that potential indicators for those viral pathogens, such as bacteria or bacteriophages, may not be as appropriate as was assumed before. For example, the NoV may be present, or persistent, in water or shellfish whereas the indicator organisms are not (Lees, 2000; Formiga-Cruz et al., 2002; Myrmel et al., 2004; Horman et al., 2004). In conclusion, the detection of NoVs in foods is still difficult, due largely to the low level, and focal way of contamination. Therefore, successful (pre-marketing) screening of foods is unlikely to be implemented soon, with the possible exception of shellfish. The generic RT-PCR protocols can, however, be optimised for sensitivedetection of one strain when sequence data for the contaminating strain is already obtained from patient diagnostics (Fig. 4.1 ). This approach was found to be successful in establishing a contaminated recreational fountain as the source of a gastroenteritis outbreak in schoolchildren by the NoV Birmingham strain (GGI.3) (Hoebe et al., 2004). Thus, for the detection of NoV in suspected foods (foods implicated by epidemiological data), optimised RT-PCR protocols may be used which might help to unambiguously link foods to outbreaks.

Fig. 4.1.

Scheme for foodborne virus tracking. Virus detection in patients and possible vectors such as food or water, to determine common source outbreaks.

While methods have worked reasonably well with artificially contaminated food items, assays applied to suspected food items collected in outbreaks have rarely yielded a positive result. Routine monitoring of foods for bacterial pathogens is itself acknowledged to be an ineffective method of assuring food safety (Motarjemi et al., 1996). The additional practical difficulties in obtaining reliable data for viral contamination mean that routine viral monitoring of foods is very unlikely to become widespread.

Next to the problem of obtaining false negative results due to practical difficulties in virus detection in foods and water there is the problem of obtaining false positive results using genome detection by PCR, i.e., the presence of viral RNA does not necessarily indicate the presence of infectious virus (Richards, 1999). Several authors have reported a poor correlation between conventional RT-PCR detection and virus infectivity in a cell culture assay: Slomka and Appleton (1998) and Duizer et al. (2004b) for FeCV after heat treatment or UV irradiation, Lewis et al. (2000) for poliovirus after UV irradiation, and Nuanualsuwan and Cliver (2002) for HAV, poliovirus and FeCV. On the other hand, for hypochlorite inactivation a good correlation was found between RNA detectability and infectivity (Nuanualsuwan and Cliver, 2002; Duizer et al., 2004b) and pretreatment (prior to RT-PCR) of heat or UV inactivated virus suspensions with RNase and proteinase K resulted in an improved correlation between PCR detectability and infectivity (Nuanualsuwan and Cliver, 2002). Recently it was shown that quantification of detectable RNA, by quantification of the RT-PCR signal or by application of quantitative RT-PCR methods, could improve the correlation too (Bhattacharya et al., 2004; Duizer et al., 2004b). However, one must bear in mind that neither PCR, nor EM, nor ELISA are viability or infectivity tests.

4.3. Virus tracking

How should all the information collected so far on the NoVs be combined to get a better understanding of their prevalence, transmission routes, or emergence? It is clear that fundamental knowledge on virus particles, sensitive and specific detection methods for the viruses in patients, foods and environmental samples and epidemiological data are all prerequisites for ‘virus tracking’. Virus tracking is a method aimed to answer questions also posed in standard outbreak investigations such as ‘which virus strain is causing this outbreak, where did it come from and how did it get into the susceptible population?’. Additionally, with virus tracking one strives to answer questions such as ‘how is this strain evolving, and why is it emerging (here and now)?’. Virus tracking is cleverly combining virological and epidemiological data, and adding molecular epidemiology to the formula. Molecular epidemiology has been defined by Janice S. Dorman as ‘a science that focuses on the contribution of potential genetic and environmental risk factors, identified at the molecular level, to the aetiology, distribution and prevention of disease within families and across populations’. Here we will describe how virus tracking of foodborne viruses was done in a European research project conducted by the ‘Foodborne Viruses in Europe (FBVE)’ network (http://www.eufoodborneviruses.net/). When fully operational, this approach will contribute to more rapid and internationally standardised assessment of the spread of foodborne viral pathogens. Mapping these pathways will allow identification of high-risk foods or processing methods, as well as high-risk import/transport routes, which subsequently can be targeted by prevention programmes.

4.3.1. An integrated molecular virological and epidemiological approach to study virus transmission

An important aspect of studying diseases internationally is having a platform in place for exchange, collection and sharing of data. Additionally, to understand modes of transmission of viruses, or to recognise a common source outbreak, combining molecular virological data with epidemiological data is essential. Key to this approach is the development of two linked and searchable databases (Fig. 4.2 ). One database aggregates harmonised epidemiological data, which can be entered in an outbreak investigation form via the Internet. To obtain that goal, a minimum dataset needs to be identified and agreed upon. Relevant questions were on the symptoms of the illness, route of transmission, setting of the outbreak, involvement of foods, and the extent of diagnostic evaluation. A web-based questionnaire is an efficient way for collection of information during outbreaks of (viral) gastroenteritis (Koopmans et al., 2003).

Fig. 4.2.

Virus tracking: linked and searchable databases.

The other database aggregates virological sequence data, and can be linked with the epidemiological data collection. Sequences can be submitted directly or – ideally – through the web-based outbreak investigation form. It is clear that for matching of sequences all participating labs need to perform RT-PCR assays that target the same or at least overlapping parts of the genome. Within the FBVE consortium the partners all used their own favourite PCR protocol for amplification of a part of the RdRp gene. This resulted in several different PCR products. There was, however, a minimal overlap of a 63 nucleotides fragment. In April 2004, the epidemiological and virological database contained 1,969 and 4,173 entries, respectively. Next to the genomic region coding for the RdRp, many sequences of the ORF2, coding for the capsid protein have been collected.

Cross reference of these two databases showed that 92% of non-bacterial gastroenteritis outbreaks were associated with NoV. Other enteric viruses detected were astrovirus, hepatitis A virus, rotavirus, and sapovirus. Overall, 10% of outbreaks were reported to be food- or waterborne. These outbreaks were significantly larger than outbreaks attributed to person-to-person transmission. A vehicle was reported in only 37% of outbreaks. In total, 69% of all the NoV outbreaks were associated with just one NoV strain, Grimsby virus. The Grimsby virus belongs to the Genogroup II.4 viruses (GGII.4, reference strain: Hu/NV/GGII/Bristol/1993/UK, other GGII.4 strains are Lordsdale virus and Camberwell virus). The proportion of Grimsby virus was highest in the winter of 2001/2002, when a novel variant Grimsby virus was detected across Europe. The Grimsby viruses were significantly more frequently detected in outbreaks labelled as person-to-person outbreaks than in food- or waterborne outbreaks, and in healthcare settings compared with other settings (FBVE network, unpublished data). In 2002, the emergence of a distinct variant GGII.4 virus made it to the headlines of the public media since this strain caused a tremendous increase in the number of outbreaks in UK hospitals and on international cruises. This event will be described in greater detail in Section 4.3.3.

4.3.2. Molecular epidemiology and virus tracking

Molecular epidemiology as it was applied in studies to transmission routes of foodborne viruses is mainly on sequence analysis of (a selected region of) the viral genome. Comparing viral genomes, at the sequence level, from samples that appear epidemiologically linked is the only way to conclusively identify common source outbreaks as such. It is clear that the accuracy of this method is dependent on the length of the fragment, the genome region, and the variation present between the virus strains under investigation. Although NoV are highly variable, and finding identical sequences in food or patient samples from different outbreaks indicates that these may be linked, the data need to be interpreted against a background of population-based data, or – minimally – data from outbreak surveillance. For instance, in the years 1996 and 2002, an epidemic wave of outbreaks occurred over a wide geographic region, in which a single variant NoV belonging to GGII.4 emerged. In these periods, finding identical sequences within the GGII.4 genogroup in different outbreaks meant nothing, other than that it was apparently an outbreak involved in the epidemic. On the other hand, in the same years, having identical sequences belonging to any of the other genotypes provided strong evidence for an epidemiological link.

Based on the detailed data from NoV typing one might argue that it is not the NoV in general that are an emerging foodborne pathogen. In fact, the NoV have been infecting people for quite a long time already. What we notice, however, is that new NoV strains emerge regularly.

4.3.3. Mechanisms of emergence: examples from FBVE

When we consider emergence of foodborne viruses, two intriguing observations were made. In the winter of 2000/2001, a novel variant NoV, designated GGIIb was identified. This strain was observed first in August 2000 in association with a large drinking-water related outbreak of gastroenteritis in France, and in the subsequent winter season in six other countries throughout Europe. The initial international outbreak occurred in three countries in association with consumption of imported oysters from France that met all microbiological quality assurance parameters. The GGIIb variant caused between 7 and 71% of all recognised NoV outbreaks detected during that season in different countries. Further characterisation of the new lineage of viruses showed a remarkable finding, namely that the newly recognised ORF1 region was associated with four different capsids. These capsids were similar to previously recognised NoVs. The conclusion was that the detected lineages in these series of outbreaks were in fact new NoV variants, generated through recombination.

Not enough is known on the immunity to NoV infections to speculate on how an existing capsid can be advantageous for a virus lineage to allow emergence based on such a recombination event. Note that viruses are obligate intracellular parasites and that recombination can occur only when two virus strains are infecting one cell at the same time. In the oyster-associated outbreaks, multiple variants were detected within the same product. Eating oysters (or other filter-feeding shellfish) contaminated with multiple strains, or ingesting multiple-contaminated products may lead to co-infection of people with related viruses, thereby facilitating the process of virus recombination (Fig. 4.3 ). This example clearly demonstrates that foodborne and waterborne transmission may have serious impact on the spread – and possibly even the generation – of emerging viral infections across countries. It also illustrates the need for virus-specific quality control criteria for food.

Fig. 4.3.

Recombination of viruses requires a double infection: two viruses in one cell.

The second observation was made in the winter of 2001/2002 when the emergence of a new GGII.4 virus occurred. Throughout Europe an increase in NoV outbreaks was noted. Sequence data of the detected viruses showed that the increase coincided with the emergence of a new GGII.4 strain. The new virus which emerged in January 2002 had a distinct sequence in the gene coding for the RdRp, had not been seen previously and was the dominant cause of NoV outbreaks by mid summer in all but one of the countries participating in the FBVE network (Lopman et al., 2004). Further research is needed to understand if and how the observed changes translate to distinct biological properties such as infectivity, antigenicity or (environmental) stability of the new strain. The example does show that the network approach used is suited to detect and investigate the emergence of a new strain. Interestingly, the strain correlated to the increase in NoV outbreaks in 2002 is related to the strains that emerged in the winter of 1995-1996, and again in 2004-2005 In all cases a global epidemic of No V outbreaks was observed. This suggests that viruses of the GGII.4 genotype have properties facilitating transmission or infection, and thereby have the propensity to cause epidemics.

Within a relatively short period of three years we have found that emergence of NoV is not just the closing of a diagnostic deficit, but we have found examples of two modes of true emergence. One way was the formation of new strains by recombination of existing strains. The other was the genetic drift of one GGII.4 strain to another new GGII.4 strain. Recombination and genetic drift, are examples of the two ways by which a single stranded RNA virus with an unsegmented genome can evolve. Both mechanisms contribute to the capacity of RNA virus to evolve rapidly and count as explanation for the observation that many of the emerging pathogens are RNA viruses.

4.4. Transmission routes

De Wit et al., (2003) report as risk factors for NoV infection: having a household member with gastroenteritis, contact with a person with GE outside the household (symptomatic shedding) and poor food handling hygiene. Other studies have shown that the eating of raw or undercooked shellfish is a risk factor (Kingsley et al., 2002b; LeGuyader et al., 2003; Doyle et al., 2004; Myrmel et al., 2004; Butt et al., 2004). The efficient, aerosolised spread of infectious viruses by vomiting, worsened by the environmental survival of NoV, contributes to efficient transmission. Another factor in the spreading of NoV is the shedding of infectious viruses by asymptomatic (and pre- or postsymptomatic) carriers and the employment of NoV shedders in the food chain.

4.4.1. Person to person

As mentioned before, NoV are transmitted by the faecal-oral route and the most important way of spreading is from person to person. If the virus is introduced in a sensitive population by a symptomatic shedder, the origin of the virus is clear. However, asymptomatic, pre- or post-symptomatic shedders, and food or water may also introduce the viruses. In addition, prolonged outbreaks may occur following environmental contamination and persistence of NoV. In these cases the origin of the virus is less clear.

Since viruses do not replicate in foods, and zoonotic transmission of NoV has not been found so far, all significant transmission routes are presumed to be a variation on person-to-person transmission. This means that NoV can be passed on from a shedder to a food item to the next victim, or water can be contaminated by a shedder and used for direct consumption or used for irrigation or washing of foods. It also implies that all foods handled by a food handler might become contaminated. But even if the source of contamination will always be a person, infections with food or water as vector for NoV transmission are referred to as food- or waterborne infections.

4.4.2. Food- and waterborne transmission

The most notorious food with respect to viral infections are the filter feeding shellfish (oysters, mussels, clams). These sea animals actively accumulate viruses from water contaminated with human excrements. The viruses are concentrated in the intestinal tract of the shellfish and may remain infectious for weeks in the natural environment of the shellfish or when stored cooled. Moreover, depuration is not an effective method to reduce the viral load as it is for reducing bacterial contamination (Schwab et al., 1998; Muniain-Mujika et al., 2002; Kingsley and Richards, 2003) and the composition of the bodymass (proteins, fat, sugars) of shellfish contributes to increased virus stability in heat processing (Croci et al., 1999). Matters are further complicated by the fact that viral contamination of shellfish is reported for batches that meet bacteriological safety standards and the absence of indicator organisms in growing waters or marketed batches is not always a guarantee for the absence of infectious NoV (Chalmers and McMillan, 1995; Schwab et al., 1998; LeGuyader et al., 2003; Nishida et al., 2003). Besides shellfish, many other food items have been implicated in NoV outbreaks: ice (Khan et al., 1994), raspberries (Ponka et al., 1999), salad vegetables, poultry, red meat, fruit, soups, desserts, savoury snacks (Lopman et al., 2003), sandwiches (Parashar et al., 1998; Daniels et al., 2000).

As mentioned before, all foods handled by an infectious food handler might be contaminated. Foods can be contaminated anywhere in the food chain, from farm to fork. The items posing the highest risk will be the products that are eaten raw or eaten without any (decontaminating) treatment. This explains why fruits and vegetables have been implicated in outbreaks, especially when harvested under conditions of poor sanitary hygiene or when irrigated with polluted water. The largest outbreak of NoV associated gastroenteritis in Australia involved over 3,000 individuals who had consumed orange juice (Fleet et al., 2000). Although no virus could be detected in the orange juice, the outbreak terminated when the juice was no longer distributed. Several areas where contamination of the juice could have occurred were identified in the production facilities. A high-risk practice is catering. Catered products are by definition products that are manually handled by a food handler at the end of the food chain, and therefore might pose a risk.

4.4.3. Airborne transmission

Airborne transmission of NoV was also noted in several outbreaks (Sawyer et al., 1988; Chadwick et al., 1994; Marx et al., 1999). For example, airborne spread of NoV and infection by inhalation with subsequent ingestion of virus particles was described for an outbreak in a hotel restaurant (Marks et al., 2000). An outbreak in a school showed evidence of direct infection by aerosolised viruses after a case of vomiting (Marks et al., 2003). In that same study it was found that cleaning with a quaternary ammonium preparation was ineffective. In all cases, viruses were spread by acts of vomiting.

4.4.4. Zoonotic transmission

The NoV that have been found in cattle (e.g. Jena and Newbury agent) and pigs so far are genetically distinct from the currently identified human NoVs (Sugieda et al., 1998; Liu et al., 1999; Dastjerdi et al., 1999; van der Poel et al., 2000, 2003; Sugieda and Nakajima, 2002; Deng et al., 2003; Oliver et al., 2003; Smiley et al., 2003; Wise et al., 2004). Whereas information on the swine enteric caliciviruses is still very scarce, recent studies on the bovine enteric caliciviruses (BEC) suggest that most BECs are best classified into a distinct genogroup (GGIII), and some may even be in a distinct genus (Oliver et al., 2003; Smiley et al., 2003; Wise et al., 2004). Nevertheless, genetic distances between animal NoV and human NoV are within the range found for different lineages of human viruses. Additionally, the animal NoV are enteric caliciviruses and possess a similar tissue tropism in their hosts as do the human NoV (Smiley et al., 2002). This suggests that given the right circumstances, interspecies transmission might occur.

A recent study of the seroprevalence in the general population and a cohort of veterinarians specialised in bovines shows IgG reactivity to recombinant bovine NoV capsid protein in 20% of controls and 28% of the vets. Since cross reactivity could not explain the difference in seroprevalence between these groups, these data indicate that infections of humans by bovine NoV have occurred (Widdowson et al., 2005). However, to date, no bovine NoV strains have been found in human infections, nor have human NoV been found to have caused infection in cattle or pigs.

The host range of NoV has been found to be very restricted, as many experimental animals have been inoculated with NoV without developing illness (reviewed by Green et al., 2001). In experimental settings only chimpanzees and macaques were found to seroconvert after inoculation with NoV (Green et al., 2001; Subekti et al., 2002; Rockx et al., 2005b). Recent studies of ‘natural’ infections showed NoV antibody prevalence in mangabey, pigtail and rhesus monkeys and chimpanzees in the USA (Jiang et al., 2004) but not in the Netherlands (Rockx et al., 2005a).

These data indicate that interspecies transmission cannot be excluded and thus there might be an animal reservoir for NoV. It is also possible that cattle might play a role as mixing vessel for recombination of NoV strains. While there currently is no proof of such a scenario, the genetic flexibility of RNA viruses is such that it should not be deemed impossible. On the other hand, the existing data do suggest that it is not a frequent event, nor that it has led to the introduction of new NoV strains that spread easily in the human population.

4.5. Prevention and control

Prevention of NoV infection including foodborne transmission will rely on high standards for personal and environmental hygiene. The goal should be to prevent contamination of food items rather than to rely on treatment processes to inactivate the viruses once present in the foods or water. The only way to achieve that will depend on increased awareness of viral food safety throughout the food chain, in particular implementation of Good Agricultural Practice, Good Manufacturing Practice and Good Hygienic Practice during food preparation. Application of HACCP can further enhance preventive measures by identifying and reinforcing the implementation of specific control measures.

To reduce the number of viral foodborne outbreaks in the future, governments should include considerations regarding viruses in the microbial food safety guidelines. In agriculture, primary products must be protected from contamination by human, animal, and agricultural wastes. It should be noted that the currently operating sewage treatment systems do not provide effluents safe from viruses (Lodder et al., 1999; Kukavica-Ibrulj et al., 2003; (Le Cann et al, 2004). Water used for the cultivation, preparation or packing of food should therefore be of controlled quality to prevent the introduction of virally contaminated water into the foodchain. Also, guidelines specifically aimed at reduction of viral contamination are needed, as it has become clear that the current indicators for, for example, water and shellfish quality are insufficient as predictors of viral contamination.

Furthermore, awareness among food-handlers, including seasonal workers, on the transmission of enteric viruses is needed (including the spread of viruses by vomiting), with special emphasis on the risk of transmission by asymptomatically infected persons and those continuing to shed virus following recovery from illness. This implies that food handlers who have increased risk of being NoV shedders have to be excluded from contact with food. Strict personal hygiene will not only reduce the number of virus introductions but also reduce the size of the outbreaks by minimising secondary transmission (Koopmans and Duizer, 2004).

4.6. Inactivation of caliciviruses

The resistance of the NoVs to be cultured (Duizer et al., 2004a) has hampered the development of reliable methods for their detection and viability testing. Therefore, knowledge of efficient inactivation methods and effective intervention in transmission pathways is limited and mostly based on studies using model viruses (reviewed in Koopmans and Duizer, 2004) or quantitative RT-PCR methods to asses RNA disintegration of NoV (Nuanualsuwan and Cliver, 2002; Duizer et al., 2004b). The most commonly used model for NoV in inactivation studies is the feline calicivirus (FeCV) and this virus is proposed as surrogate virus for testing of virucidal activity of disinfectants (Steinmann, 2004). In several tests it was found that FeCV was not efficiently inactivated on environmental surfaces or in suspension by, for example, 1% anionic detergents, quaternary ammonium (1:10), hypochlorite solutions with < 300 ppm free chlorine, or less than 50% or more than 80% alcohol preparations (ethanol or 1- and 2-propanol) (Scott, 1980; Gehrke et al., 2004; Duizer et al., 2004a). Varying efficacies of 65 to 75% alcohol preparations are reported, however, short contact times (< 1 min) rarely resulted in more than 4-log inactivation. Moreover, the presence of faecal or other organic material reduces the virucidal efficacy of many chemicals tested. The efficacy of seven commercial disinfectants for the inactivation of FeCV on strawberries and lettuce was tested by Gulati and co-workers and they found that none of the disinfectants was effective (defined as reducing the virus titer by at least 3-log10) when used at the FDA permitted concentration (Gulati et al., 2001). These data were recently confirmed by Alwood and co-workers for sodium bicarbonate, chlorine bleach, peroxyacetic acid and hydrogen peroxide at FDA approved concentrations (Allwood et al., 2004).

Recent studies using FeCV and canine caliciviruses (CaCV) as models for NoV showed that heat inactivation of these two animal caliciviruses was highly comparable. Based on the temperature dependent inactivation profiles it was suggested that a better inactivation of viruses may be expected from regular batch (63 °C for 30 min) or classical pasteurisation (70 °C for 2 min) than from high temperature short time (HTST) pasteurisation (72 °C for 15 sec). Inactivation at 100 °C was, however, complete within seconds (Duizer et al., 2004a).

The resistance of FeCV (in suspension) to inactivation by UV 253.7 nm radiation was reported to be highly variable. Doses required to achieve 3-log10 reduction in FeCV infectivity ranged from 12 to 26 mJ/cm² (De Roda Husman et al., 2004; Thurston-Enriquez et al., 2003), while others found only 1-log10 reduction at 48 mJ/cm² (Nuanualsuwan et al., 2002). These differences may result from differences in composition (turbidity) of the irradiated suspensions although it was shown that at low protein levels (< 4 ug/ml) the effect of suspension composition was negligible.

From two studies comparing FeCV and CaCV it may be concluded that UV-B radiation is less effective than UV 253.7 nm in inactivating caliciviruses in suspension (De Roda Husman et al., 2004; Duizer et al., 2004a). The relative resistance of FeCV to UV radiation is intermediate, i.e., comparable to that for the enteroviruses (Gerba et al., 2002), less effective than for vegetative bacteria, but more effective than for phage MS2 (De Roda Husman et al., 2004) and for adenoviruses 2 and 40 (Gerba et al., 2002; Thurston-Enriquez et al., 2003) and B. subtillis spores (Chang et al., 1985). The inactivation of caliciviruses by ionising (gamma) radiation was found to be ineffective, requiring doses of 300-500 Gy to achieve 3 log reduction, i.e., more than twice as much as for phage MS2. Moreover, the low efficacy was even further reduced by increasing the protein content of the suspension to 3-4 ug/ml (De Roda Husman et al., 2004). Other studies reported no apparent effect of ultrasonic energy (26 kHz) on FeCV (Scherba et al., 1991), but complete inactivation by high hydrostatic pressure after 5-min treatments with 275 MPa or more (Kingsley et al., 2002a). Based on comparative PCR data it was concluded that FeCV was a valuable model for NoV in inactivation experiments. However, the enteric NoV was significantly less sensitive to low pH treatments, indicating the need for a truly enteric virus as model for NoV or for an in vitro method for the detection of NoV viability (Duizer et al., 2004b).

Due to the high stability of NoV they can survive in the environment, in water or on foods for prolonged periods. This means that the contamination can have occurred a long time ago. Additionally the NoV are quite resistant to refrigeration and freezing, low pH and chemical disinfectants. This combination of high stability and resistance with a low minimal infectious dose has led to several remarkable outbreaks as described above. Awaiting further data on inactivation of NoV by milder disinfectants or treatments it is probably best to apply thorough heating (cooking) for water and food, and high hypochlorite concentrations (> 1%) for disinfection of surfaces when contamination with NoV is suspected.

4.7. Thoughts on other viruses

We have learned a great deal on the transmission routes of NoVs due to the fact that infections due to these viruses are relatively easily recognised. The attack rate is high, and the symptoms develop fast, resulting in outbreaks that are easily recognised. The work of the FBVE network has shown that foodborne transmission of viruses contributes significantly to the epidemiology and disease burden of NoV. What does that tell us about food-related risks for other viruses, especially the structurally similar viruses, such as enteroviruses (e.g. poliovirus, coxsackie viruses), hepatitis A virus, and hepatitis E virus? The basic properties that contribute to foodborne transmission, such as asymptomatic carriage and shedding, environmental stability, oral infectivity, and low minimal infectious dose are similar to those of the NoVs. Several examples of food- and water-related outbreaks with these viruses exist. The biggest difference with NoVs is that enteroviruses, hepatitis A and hepatitis E viruses cause illness in a smaller proportion of those exposed, and that these symptoms become noticeable up to ten weeks post infection. As a result, a relation between the illness and a food source will not be made unless it is very obvious, for instance, because a group of cases all participated in the same conference. Otherwise, such outbreaks can be detected only by the use of virus tracking, as described above.

The data described also illustrated that current microbiological quality assurance is not adequate for effective prevention of virus infection. This is an important message, because the globalisation of the food chain and international travel are contributing factors to the rapid dissemination of emerging diseases. This is illustrated by the example of SARS, which was considered to be a respiratory pathogen, where control efforts have focused on control of droplet transmission in close person-to-person contacts. Epidemiological and laboratory studies have, however, shown that a major proportion of the SARS cases in Hong Kong has been linked to environmental transmission due to a faulty sewage system (WHO, 2003). Moreover, SARS coronavirus is found to infect intestinal epithelial cells in humans and the majority of SARS patients shed high loads of infectious virus in their stools (Leung et al., 2003), SARS coronavirus is relatively stable in the environment (Duan et al., 2003), coronaviruses in animals are known for their dual tissue tropism (respiratory and enteric), and food animal handlers in Guangdong had increased prevalence of antibodies against the SARS virus (CDC, 2003). Thus, with increasing data becoming available, one can question if the droplet-based respiratory transmission is truly the only mode of transmission. All the factors mentioned suggest that the potential of transmission of SARS coronavirus via human faecal waste is there. During the major outbreak in 2003, foodborne transmission has not been documented, but the latest SARS cases in Guangdong that occurred in December/January 2004 were linked to two restaurants. Without being alarmed, these factors should be used as a warning sign. If SARS is transmitted via food, the NoV data show that we do not have sufficient safeguards in place to prevent human disease.

4.8. Future trends

Given the importance of foodborne viruses and the impact that different factors (globalisation of the market, increased international travel, consumer demands, changes in food-processing, pathogen evolution, etc.), may have on the emergence of disease, it is clear that priority needs to be given to expanding foodborne disease surveillance to cover foodborne viruses. The expansion should include three areas: (i) a more complete coverage of qualified laboratories per country, (ii) inclusion of more countries in international surveillance networks and (iii) development and implementation of detection methods of more classes of viruses in the surveillance programs. Networks comparable to the FBVE network are being developed in other regions of the world and it is important to develop guidelines and rules for data sharing in these early stages.

It is also worthwhile paying attention to the impact on viruses of new mild preservation techniques used to inactivate bacteria in foods or, for example, alcohol-based hand-sanitisers used to inactivate bacteria on hands. Most of these methods have been introduced and put into use without proper evaluation of the efficiency against viruses, creating a window of opportunity for foodborne viruses to cause problems.

It is also to be expected that the attention given to hepatitis E virus (HEV) will expand significantly since the recent proof of foodborne zoonotic transmission of these viruses (Tei et al., 2003; Yazaki et al., 2003) (HEV will be discussed in detail in Chapter 11). Increased awareness of food as a vector for viruses, combined with increased laboratory capabilities and the implementation of virus tracking methods and molecular epidemiology might result in detection of foodborne transmission of as yet unforeseen viruses. Hopefully the diagnostic gap in gastrointestinal illnesses such as gastroenteritis and hepatitis will be narrowed down. It is not very likely, however, that the problems caused by foodborne virus transmission are going to be resolved quickly. Continued research is needed to assess efficacy of different control measures along food chains to be able to work towards virus-safe food in the future.

4.9. Additional sources of information

Foodborne viruses: An Emerging Problem. Ilsi report and Int J Food Microbiol. 2004 Jan 1; 90(1): 23-41. Review.

Foodborne viruses. Koopmans M, von Bonsdorff CH, Vinje J, de Medici D, Monroe S. FEMS Microbiol Rev. 2002 Jun; 26(2): 187-205. Review.

Viruses. Koopmans M. (2002) in Foodborne Pathogens: hazards, risk analysis and control, Eds Blackburn and McClure. CRC press.

Human Caliciviruses. Kim Y. Green, Robert M. Chanock and Albert Z. Kapikian. In Fields Virology (4th edn, 2001), vol. 1, 841-874.

Molecular epidemiology of human enteric caliciviruses in The Netherlands. Koopmans M, Vinje J, Duizer E, de Wit M, van Duijnhoven Y. Novartis Found Symp. 2001; 238: 197-214; discussion 214-8. Review. http://www.eufoodborneviruses.net/