A Review of Known and Hypothetical Transmission Routes for Noroviruses

Abstract

Human noroviruses (NoVs) are considered a worldwide leading cause of acute non-bacterial gastroenteritis. Due to a combination of prolonged shedding of high virus levels in feces, virus particle shedding during asymptomatic infections, and a high environmental persistence, NoVs are easily transmitted pathogens. Norovirus (NoV) outbreaks have often been reported and tend to affect a lot of people. NoV is spread via feces and vomit, but this NoV spread can occur through several transmission routes. While person-to-person transmission is without a doubt the dominant transmission route, human infective NoV outbreaks are often initiated by contaminated food or water. Zoonotic transmission of NoV has been investigated, but has thus far not been demonstrated. The presented review aims to give an overview of these NoV transmission routes. Regarding NoV person-to-person transmission, the NoV GII.4 genotype is discussed in the current review as it has been very successful for several decades but reasons for its success have only recently been suggested. Both pre-harvest and post-harvest contamination of food products can lead to NoV food borne illness. Pre-harvest contamination of food products mainly occurs via contact with polluted irrigation water in case of fresh produce or with contaminated harvesting water in case of bivalve molluscan shellfish. On the other hand, an infected food handler is considered as a major cause of post-harvest contamination of food products. Both transmission routes are reviewed by a summary of described NoV food borne outbreaks between 2000 and 2010. A third NoV transmission route occurs via water and the spread of NoV via river water, ground water, and surface water is reviewed. Finally, although zoonotic transmission remains hypothetical, a summary on the bovine and porcine NoV presence observed in animals is given and the presence of human infective NoV in animals is discussed.

Introduction

Noroviruses (NoVs) were discovered in 1972 by immune electron microscopy analysis of fecal samples of volunteers challenged with fecal filtrates from a group of elementary school students affected by an outbreak of gastroenteritis in 1968 in Norwalk, Ohio (Kapikian et al. 1972). The 27–32 nm viral agent was originally named Norwalk virus and was later recognized as the type agent of the genus Norovirus (previously denoted as “Norwalk-like viruses” or “small round structured viruses”). Similar to human infective NoVs, animal infective NoVs were discovered by electron microscopy of stool samples of domestic animal species (calves and pigs) suffering from gastroenteritis (Bridger 1980; Saif et al. 1980; Woode and Bridger 1978).

Together with the genera Sapovirus (previously called “Sapporo-like viruses”), Lagovirus, Vesivirus, and Nebovirus, the Norovirus genus forms the Caliciviridae family (Green et al. 2000). The Norovirus genus comprises five genogroups (GI to GV) each containing several NoV genotypes (Zheng et al. 2006). Genogroups I and II (GI and GII) consist of 8 and 17—extended to 19 by Wang et al. (2007)—–NoV genotypes, respectively. These genogroups comprise the human infective NoV genotypes, together with the Alphatron and Ft. Lauerdale genotypes in genogroup IV (GIV). The latter genogroup also contains a number of NoV genotypes infecting carnivores such as dogs, lions, and cats (Martella et al. 2007; Martella et al. 2008; Pinto et al. 2012). Bovine and murine NoV are classified in genogroup III (GIII) and V (GV), respectively, while porcine NoV is also classified in GII. Human infective NoVs are among the most important causes of gastroenteritis in adults worldwide and NoV infections often occur as outbreaks (Koopmans and Duizer 2004). On the other hand, the true extent of the host range of animal NoVs remains unknown although it is known that a large number of animal species can be infected with NoVs (Scipioni et al. 2008).

Human and animal infective NoVs are spread via feces and vomit through different transmission routes. Due to the development of sensitive methods for molecular detection of NoVs, knowledge on these NoV transmission routes has increased substantially. The current review, therefore, aimed to provide an overview of known transmission routes, specifically of human infective NoVs. Furthermore, the presence of animal infective NoVs in bovine and porcine animals and the possibility of zoonotic transmission of NoVs were discussed.

Transmission Routes of NoV

Described transmission routes of human infective NoVs include person-to-person transmission (chapter 3.1), food borne transmission (chapter 3.2), and water borne transmission (chapter 3.3) while zoonotic transmission (chapter 3.4) has been considered a hypothetical—yet unproved—route for NoV transmission.

A schematic overview of proven and hypothetical transmission routes for human and animal infective NoVs is shown in Fig. 1. Transmission of human infective NoVs is facilitated by a number of factors such as (i) prolonged duration of viral shedding, even after resolving of the symptoms (Aoki et al. 2010; Atmar et al. 2008; Lee et al. 2007), (ii) fecal shedding during asymptomatic NoV infections (Gallimore et al. 2004b; Ozawa et al. 2007; Phillips et al. 2010), and (iii) NoV persistence on environmental surfaces and in water and foods (Escudero et al. 2012; Lamhoujeb et al. 2009; Liu et al. 2009b; Ueki et al. 2007).

Fig. 1

Schematic overview of the transmission routes of human and animal infective NoVs. Solid and dashed arrows indicate proven and hypothetical transmission routes, respectively. The thickness of the arrows is related to the likeliness of the transmission route

Person-to-Person Transmission

Although human infective NoVs can be transmitted through several routes, person-to-person transmission is considered to be the dominant NoV transmission route (Fig. 1) (Kroneman et al. 2008). NoV outbreaks related to this transmission route have extensively been documented in semi-closed community settings such as hospitals, cruise ships, day-care centers, and military settings and can affect a few to a lot of people (Gallimore et al. 2004a, b; Grotto et al. 2004; Takkinen 2006; Wick 2012). Recently, NoV outbreaks between 2002 and 2006 were investigated using data collected by 11 surveillance systems of the Food borne Viruses in Europe (FBVE) network (Verhoef et al. 2010). Out of 886 confirmed norovirus outbreaks with known transmission route, 654 outbreaks (74 %) involved person-to-person contamination. Similarly, Lopman et al. (2003) and Siebenga et al. (2007) found that person-to-person spread was the transmission route in 85 % and 82 % of NoV outbreaks with a known mode of transmission in the UK and in The Netherlands, respectively.

The NoV genotype most commonly identified in person-to-person NoV gastroenteritis outbreaks is the NoV GII.4 genotype (Kroneman et al. 2008; Verhoef et al. 2010, 2009). Therefore, its epochal evolution has been documented thoroughly between 1974 and 2007. In the USA, only 2 major NoV GII.4 strains were observed between 1974 and 1994 (a NoV GII.4 ancestor strain and GII.4 Camberwell). In the 7 year period between 1995 and 2002 a single major NoV GII.4 strain was considered dominant (NoV GII.4 Grimsby) both in Europe and in the USA. Subsequently, NoV GII.4 Farmington Hills and Hunter (also named NoV GII.4 2002 and NoV GII.4 2004, respectively, in Europe) caused 2 NoV epidemic seasons in 2002–2003 and 2004–2005. Finally, NoV GII.4 Laurens and NoV GII.4 Den Haag (also named GII.4 2006a and GII.4 2006b in Europe) have caused NoV epidemics since 2006 (Bok et al. 2009; Siebenga et al. 2009; Zheng et al. 2009). Bull and White (2011) recently reviewed reasons for the success of the NoV GII.4 genotype and they concluded that a synergism of multiple factors could be responsible for its success. First of all, NoV GII.4 has a mutation rate that is approximately 2-fold higher compared to other NoV genotypes (Bok et al. 2009; Bull et al. 2010; Siebenga et al. 2010). This could in part be caused by the NoV GII.4 RNA dependent RNA polymerase which has shown a lower fidelity compared to some other NoV genotypes, enabling the more prevalent viruses to avoid immune recognition by rapidly altering their antigenic properties (Bull et al. 2010; Bull and White 2011). Second, NoV GII.4 can bind more histo-blood group antigen (HBGA) types than any other NoV genotype, which could be a major contributing factor to the NoV GII.4 dominance (Bull and White 2011). Third, while long-term immunity may be possible for some NoV genotypes (e.g., GII.3), it is possible that this is not the case for GII.4 NoV (Bull and White 2011; Siebenga et al. 2009; Yang et al. 2010). However, only limited research has been performed on each factor and further study is needed.

Food Borne Transmission

Person-to-person transmission is without a doubt the dominant transmission route for NoV, but the primary cases in NoV outbreaks often have a food or water borne cause (Fig. 1). Person-to-person transmission among contacts of primary cases can further propagate the epidemic (Becker et al. 2000; Patel et al. 2009). Food products can be contaminated with NoV by contact with fecal material or vomit, which can occur during any stage of the food production (Baert et al. 2011). In essence, NoV contamination of foods can occur during at a pre-harvest level, e.g., by irrigation of fresh produce with NoV contaminated water or by use of contaminated manure, or at a (post-) harvest level, e.g., by manual picking, processing, and preparation of foods such as strawberries and raspberries (Wei and Kniel 2010; Zainazor et al. 2010).

The current paragraph will go more into detail regarding these transmission routes by a literature review of studies investigating individual NoV food borne outbreaks. In total, 51 studies describing 58 outbreaks that occurred between 2000 and 2010 were reviewed (Table 1). The number of described NoV food borne outbreaks is considered as a serious underestimation of the actual number and is probably biased because peer-reviewed publications of food borne outbreaks merely report large, well-documented, unusual, or novel events (Baert et al. 2009b; Kroneman et al. 2008; O’Brien et al. 2006). Furthermore, NoV illness is normally self-limiting and complications are not common which could lead to a further underestimation of NoV food borne gastroenteritis outbreaks (Greening 2006; Lopman et al. 2002).

Table 1 Overview of NoV food borne outbreaks between 2000 and 2010

A food borne outbreak has been defined by multiple official authorities such as the European Food Safety Authority (EFSA), the World Health Organization (WHO), and the Centers for Disease Control and Prevention (CDC). However, all definitions include the occurrence of two or more human cases of a same/similar disease resulting from the same food source(s) (Anonymous 2000, 2010; World Health Organization 2008). All reviewed outbreaks complied with these definitions and the NoV food borne outbreak data were categorized in Table 1 per transmission route: by pre-harvest contamination or by (post-) harvest contamination. Regarding the latter transmission route, focus was set on infected food handlers, either confirmed or suspected (Table 1). The involvement of a NoV infected food handler in a NoV food borne outbreak was considered confirmed if (i) epidemiological analysis showed that the food handler was infected before the food borne outbreak and could be linked to manipulation of the involved food products and if (ii) laboratory analysis showed identical NoV genogroups or genotypes in the clinical samples from both patients and food handler. For every reviewed NoV food borne outbreak in Table 1, the attack rate (the number of exposed individuals infected with NoV divided by the number of exposed individuals in a NoV food borne outbreak), and/or the number of affected people and the laboratory confirmation of the NoV presence in human and/or food samples were described. Regardless of the transmission route, some trends were clearly noticeable. Overall, RT-PCR was the most used NoV detection method as it was used in 47 of 51 (92 %) studies. Extraction of virus particles or viral genomic material from the food samples was performed by various methods. Of the 23 studies (45 %) that tested the NoV presence on food products, 14 studies (61 %) used elution-concentration extraction, proteinase K treatment, and direct RNA extraction to extract NoV from food, while 9 (39 %) studies did not specify the used virus extraction method (Table 1). In general, all but two studies (Boxman et al. 2007; Kingsley et al. 2002) combined epidemiological investigations with molecular detection of NoVs in clinical samples. A smaller fraction of studies (29 %) was able to demonstrate the NoV presence in the suspected food products (Table 1).

Pre-Harvest Contamination of Food Products

In general, food products most at risk for pre-harvest NoV contamination include fresh produce and shellfish (Baert et al. 2011; Lowther et al. 2012; Mattison et al. 2010; Stals et al. 2011b). NoV contamination of fresh produce at a pre-harvest stage can result from contact with polluted irrigation water or contaminated manure (Wei and Kniel 2010), while shellfish can be contaminated if grown in NoV contaminated harvesting areas (Lowther et al. 2008). Table 1 showed that 24 out of 58 (41.4 %) NoV food borne outbreaks could be related to consumption of pre-harvest contaminated fresh produce and shellfish and that mainly raspberries and oysters were involved in these outbreaks. An average attack rate of 51 % was observed with an average of 145 people exposed per outbreak. The most observed NoV genotype in clinical and food samples was GII.4, although NoV genotypes GI.1, GI.2, GI.4, and GII.7 were frequently detected in food and clinical samples as well.

In 42 % (10 out of 24) of these outbreaks, identical genogroups or genotypes were found in clinical and food samples. In an additional 8 % (2 out of 24) of these NoV food borne outbreaks, different NoV genotypes were detected in clinical and food samples. Noteworthy, NoV could be detected in shellfish samples in 9 of 11 (82 %) studies, which could be explained by the higher levels of NoV found in shellfish compared to other food products due to their filter feeding capacity (Meyers 1984).

NoV and other viral pathogens have been observed in shellfish by several screening studies. A 3 year survey between 2005 and 2008 investigating 116 retail shellfish samples (mussels, clams, and oysters) showed a confirmed (sequenced) detection of NoV genotypes GII.4 2004 and GIIb in 10.3 % of all tested samples (Terio et al. 2010). Likewise, GII.4 and GIIb genotypes were found in 16.7 % of oyster and mussel samples (n = 42) during a 2 year survey for the NoV presence in Dutch shellfish (Boxman et al. 2006) while a 1 year survey in 235 Italian shellfish samples showed the presence of NoV and HAV in 13.2 % and 2.2 % of all tested samples, respectively (Croci et al. 2007). Although some studies found NoV GII.4 in shellfish samples related to NoV food borne outbreaks, the GIIb genotype was not encountered in any of these studies (Table 1). A lower NoV presence was observed when 1512 Japanese oysters were screened for the NoV presence as a broad range of GI and GII NoV genotypes were found in 4.9 % of all tested oyster samples (Nishida et al. 2007). NoV have been detected in sewage with high concentrations (339 to 10⁶ NoV genomic copies per liter) and treatment of the sewage caused only a minor reduction of 0.7–2.7 logs NoV genomic copies per liter (Lodder and de Roda Husman 2005; van den Berg et al. 2005). A Japanese study has shown that very similar NoV genotypes can be detected in human feces, domestic sewage, treated wastewater, river water, and in cultivated oysters (Ueki et al. 2005), demonstrating that transmission of NoV from contaminated harvesting water to bivalve filter feeding shellfish such as mussels or oysters is possible. Although numerous studies using PCR have demonstrated the NoV presence in various shellfish and shellfish harvesting areas, the number of epidemiologically confirmed shellfish-associated outbreaks is relatively low. Lowther et al. (2012) found that the number of genomic copies detected in shellfish samples was significantly higher in outbreak related samples (mean level of 2148 NoV genomic copies/g) compared to non-outbreak related samples (mean level of 682 NoV genomic copies/g). Although more data is needed, a critical level may aid NoV risk management in shellfish. As critical limits of 100, 200, 500, 1000, or 10.000 NoV genomic copies would result in the rejection of 33.6 to 88.9, 24.4 to 83.3, 10.0 to 72.2, 7.7 to 44.4, or 0 to 11.1 % of batches, respectively (Anonymous 2012), careful consideration of a possible NoV critical limit in oysters and other shellfish is needed.

Regarding the NoV food borne outbreaks related to consumption of pre-harvest contaminated raspberries, two out of five studies were able to recover NoV from this food matrix (Le Guyader et al. 2004; Maunula et al. 2009). Multiple authors have stated that detection of NoVs in raspberries can be challenging, especially if the NoV levels are lower than 10⁴ genomic copies per 10–15 g (Stals et al. 2011c; Summa et al. 2012a). However, as a quantitative method was not applied in both outbreak studies, the NoV levels on the raspberries could not be determined. Only a limited number of studies have investigated the pre-harvest presence of enteric viruses on fresh produce. NoV and adenovirus have been detected in a single spinach sample when screening 30 produce samples for the enteric virus presence (Cheong et al. 2009). A recent review by Baert et al. (2011) summarized a Belgian, a Canadian, and a French study screening the NoV presence on fresh produce (Mattison et al. 2010; Stals et al. 2011b). NoV was detected by real-time RT-PCR in 28 % (n = 641), 33 % (n = 6), and 50 % (n = 6) of leafy greens tested in Canada, Belgium, and France, respectively. Soft red fruits were found positive in 35 % (n = 29) and 7 % (n = 150) of the samples tested in Belgium and France, respectively. However, subsequent sequencing of a conventional RT-PCR amplicon was successful in merely 7 % of all positive results. The latter finding gave rise to questions regarding the interpretation of NoV-positive real-time RT-PCR results in the light of the increase in sensitivity of the NoV detection methodology. The authors suggested that strategies to confirm the results by real-time RT-PCR should be developed in analogy with the detection of microbial pathogens in foods (Baert et al. 2011). Although a critical limit for food products such as raspberries and other fresh produce–similar to shellfish–is an option for risk management, further research is needed regarding a potential link between NoV levels on fresh produce and related food borne illness.

(Post-) Harvest Contamination of Food Products

Contamination of food products at post-harvest stage can occur at any point during harvesting, processing, preparing, and packing of the food (Moe 2008; Todd et al. 2009). The current review focussed on the transmission of NoV by infected food handlers, as they have been confirmed to play a major role in NoV transmission (Baert et al. 2009b; Widdowson et al. 2005b). Table 1 showed that food handlers were involved in 34 out of 58 (58 %) NoV food borne outbreaks, either suspected (16 out of 34 outbreaks) or confirmed (18 out of 34 outbreaks). Deli sandwiches were the most frequent implicated foods in these food borne outbreaks, but as expected, a wide range of food products including catered meals, buffet foods, and prepared salads have been involved. An average attack rate of 34 % was observed with 120 people averagely exposed per outbreak, which was comparable to the NoV food borne outbreaks caused by pre-harvest contaminated foods. Remarkably, NoV could be detected in food samples in merely 12 % (4 out of 34) of studies investigating NoV food borne outbreaks related to infected food handlers, while this was possible in 50 % (12 out of 24) of food samples in studies investigating NoV food borne outbreaks related to pre-harvest contamination of foods. This difference can be explained by the fact that NoV detection in ready-to-eat foods such as catered foods and deli sandwiches is substantially more difficult compared to fresh produce and shellfish (Stals et al. 2011a, c). The main reason for this is the composition of the food products; while fresh produce is largely composed of carbohydrates and water, ready-to-eat foods contain a lot of fat and proteins (Baert et al. 2008). As expected, the most frequently detected NoV genotype in clinical samples was the GII.4 genotype, although NoV GI.3 was detected in some outbreaks as well (Table 2).

Table 2 Overview of bovine NoVs detected worldwide

The role of the food handler is considered an important factor for food borne transmission of NoVs due to several factors. First of all, it has been stated that an apparent NoV infection of a food handler should always be reported to avoid NoV food borne outbreaks and that an infected food handler should not be at a work place where foods are manipulated (Joint FAO/WHO Codex Alimentarius Commission and Joint FAO/WHO Food Standards Program 2003; Moe 2008; Vivancos et al. 2009). Since food handlers are often related to large catering establishments, outbreaks related to this transmission route tend to affect a lot of people at once (Noda et al. 2008). However, food handlers returning to work after recovering from a NoV infection can still shed considerable NoV levels and can cause NoV food borne outbreaks (Atmar et al. 2008; Malek et al. 2009). Second, food handlers carrying an asymptomatic NoV infection can easily cause food borne outbreaks since they can shed similarly high NoV levels (Ozawa et al. 2007; Phillips et al. 2010). Asymptomatic NoV infected food handlers manipulating deli sandwiches, prepared meals, and salad vegetables have caused NoV food borne outbreaks (Godoy et al. 2005; Ohwaki et al. 2009; Vivancos et al. 2009). Third, poor personal hygiene has been reported as well (Rizzo et al. 2007; Schmid et al. 2007), often in combination with food handlers suffering from NoV illness still coming to work (Grotto et al. 2004). Fourth, lack of respect to hygienic working circumstances can contribute to NoV food borne outbreaks (Friedman et al. 2005; Schmid et al. 2007). For example, sinks used for both washing hands and washing lettuce have been related to a NoV food borne outbreak (Payne et al. 2006). Another study reported vomiting of an ill baker in a sink in the food preparation area which resulted in NoV contamination of deli sandwiches (De Wit et al. 2007). Finally, an underestimated factor may be the presence of ill children and other family members in the food handler’s home environment. Nevertheless, a few studies have mentioned this as a risk factor for NoV food borne outbreaks (Boxman et al. 2007; Daniels et al. 2000).

Food Borne Norovirus Outbreaks in Belgium (2004–2009)

As the current review has been a cooperation between different Belgian laboratories, we were able to review Belgian NoV food borne outbreaks more into detail with regards to the possible NoV transmission routes. In Belgium, a convention between the Flemish Community and the National Reference Laboratory for Food borne Outbreaks (NRL-FBO) allowed analysis by the latter laboratory of human samples in suspected NoV food borne outbreaks.

In Belgium, it has been observed that in 20–50 % of the reported food borne outbreaks the causative agent remains unknown and it is suspected that NoV could be partially responsible for these unknown cases. Reasons for this include a low reporting rate as NoV illness is normally self-limiting and complications are rare (Greening 2006; Lopman et al. 2002). Furthermore, analysis of clinical samples of patients is not reimbursed in Belgium. A NoV extraction and detection protocol as described by Baert et al. (2008) and Stals et al. (2011a) was introduced in the NRL-FBO in 2006. While only six NoV food borne outbreaks were altogether observed in 2004, 2005, and 2006, this number increased in 2007 to ten while seven NoV food borne outbreaks were reported in 2008 and in 2009.

In 2006, two subsequent NoV food borne outbreaks in a care center for disabled persons were investigated. During the first episode, 12 persons became ill and NoV GII was found in a “witness meal,” a food sample that is explicitly stored for a certain time (usually 72 h to 1 week at 4 °C or at −20 °C) to allow analysis of the food in case of a food borne outbreak. Four months later, 50 people became ill in the same institute and a mixture of NoV GI and GII was found in one of the witness meals analyzed. These subsequent outbreaks are an example of recurring NoV outbreaks in health care facilities such as nursing homes and hospitals (Cieslak et al. 2009; Koopmans 2009; Rosenthal et al. 2011). A third outbreak took place in a hospital where 17 out of 400 people became ill. NoV GII was detected in the soup and also in five out of six fecal samples and epidemiological data indicated that an infected food handler distributing the soup could have been at the origin of this infection. Although soup is not commonly associated with NoV outbreaks, it has been pointed out as a vehicle for NoV transmission in a NoV outbreak in a residential-care facility for the elderly (Medici et al. 2008).

In ten out of the 75 food borne outbreaks reported to the NRL-FBO during 2007, NoV was confirmed as causative agent and in total 392 persons were affected. The majority of outbreaks occurred at work (30 %), camp sites (20 %), and nursing homes (20 %), and the involvement of a food handler was suspected in eight outbreaks. Noteworthy, the NoV presence could be shown in food samples in five out of ten (50 %) NoV food borne outbreaks demonstrating the effectiveness of the NoV detection method described by Baert et al. (2008) and Stals et al. (2011a) on naturally NoV contaminated food samples such as soup, chicken and rice, mashed potatoes, meat stew, and a composite meal. The most important vehicles for NoV transmission were deli sandwiches as they were involved in four out of ten NoV food borne outbreaks.

NoV caused seven food borne outbreaks in 2008. In two outbreaks, deli sandwiches were indicated as vehicles for NoV transmission. While only fecal samples showed the NoV presence in a first outbreak affecting 200 people, three deli sandwich fillings (meatball, cheese, and chicken curry) and food handler’s fecal sample were found NoV positive in a second outbreak affecting 15 people. A third outbreak resulted in NoV illness of 80 people participating at different barbecues. Fecal samples from ill persons in the latter outbreak tested positive for the NoV presence, but NoV could not be found in the food. Interestingly, in the food handler’s family there was a history of gastroenteritis a week before the barbecues took place, which confirms the importance of NoV infections in the food handler’s environment (Boxman et al. 2007; Daniels et al. 2000).

Finally, in five out of seven NoV food borne outbreaks in 2009, the NoV presence could only be confirmed in fecal samples of patients and/or of food handlers. In two out of seven outbreaks, foods were confirmed as vehicle for NoV transmission.

The NoV food borne outbreaks in Belgium confirm trends that have been observed worldwide. Most importantly, the majority of NoV food borne outbreaks were related to infected food handlers, which confirms their important role in NoV transmission (Moe 2008).

Water Borne Transmission

Transmission of NoV via water can occur by several routes, although initial contamination is always caused by discharge of human fecal material (Bosch 1998). NoVs and NoV surrogate viruses can efficiently attach to lettuce veins and can internalize lettuce leafs during irrigation. The latter can lead to NoV contamination of the fresh produce (Gandhi et al. 2010; Li et al. 2012; Vega et al. 2005, 2008; Wei et al. 2010, 2011). River water, surface water, and ground water are frequently used for irrigation of crops (Knox et al. 2011; Steele and Odumeru 2004). Furthermore, people can be infected with NoV via consumption of drinking water or via contact with recreational surface water (Bosch 1998; Moe et al. 2007). As a thorough overview of water related transmission routes for NoV and other viral pathogens has been described by Bosch et al. (1998), the current review will focus specifically on NoV transmission via river water, ground water, and surface water. It should be noted that direct comparison of studies investigating the NoV presence in water is difficult, as most studies use different NoV detection methodologies.

River Water

River water is likely to be contaminated with NoV as rivers are continuously fed with effluents of wastewater treatment plants, which are optimized for the removal of bacteria while removal viral pathogens could be less efficient (da Silva et al. 2007; Hewitt et al. 2011; Maunula et al. 2012; Ueki et al. 2005). The NoV presence in river water used for irrigation and used as drinking water has extensively been investigated in numerous screening worldwide and except for a single study (La Rosa et al. 2007), the NoV presence has been shown in all tested rivers (Haramoto et al. 2005; Jurzik et al. 2010; Laverick et al. 2004; Lee and Kim 2008; Lodder et al. 2010; Mans et al. 2012; Schets et al. 2008; Westrell et al. 2006). In some cases, other enteric viruses such as rotaviruses, enteroviruses, and adenoviruses have been detected simultaneously (Haramoto et al. 2005; He et al. 2011; Jurzik et al. 2010; Kishida et al. 2012; Kiulia et al. 2010; Lodder et al. 2010; Schets et al. 2008). Regarding the NoV concentrations found in river water used for irrigation, a broad range of NoV levels can be present, fluctuating between 12 NoV genomic copies/l and 6.4 × 10⁴ NoV genomic copies/l, although most studies only indicated the presence/absence of NoV in water (Haramoto et al. 2005; Jurzik et al. 2010; Laverick et al. 2004; Lodder et al. 2010; Westrell et al. 2006). Overall, a broad range of NoV genotypes have been detected in river water. While some studies have shown the dominant presence of the GII.4 genotype (Blanco Fernandez et al. 2011; Lodder et al. 2010), most studies demonstrate the simultaneous presence of various NoV genotypes such as GI.4 and GI.5 alongside other less frequently observed NoV genotypes (La Rosa et al. 2007; Lee and Kim 2008; Mans et al. 2012). NoV levels in river water depend on whether sewage and other sources containing human fecal material is discharged and whether or not the sewage has been treated (Lodder et al. 2010; Teunis et al. 2009). Noteworthy, a seasonal variation in NoV levels has been observed in long-term studies as NoVs were in generally more frequently observed in winter and early spring (He et al. 2011; Kishida et al. 2012; Kitajima et al. 2009, 2011; Pérez-Sautu et al. 2012; Westrell et al. 2006). This seasonality is most likely related to a combination of (i) the fact that low temperatures are considered a good conservation method for viruses (Baert et al. 2009a) and (ii) higher prevalence of NoVs in the human population during winter (Ahmed 2012; Rohayem 2009).

Ground Water

The microbial quality of ground water is in general considered very good as this water type tends to be cooler, is protected from sunlight, and has less microbiological and biological activity (Feachem et al. 1983; Steele and Odumeru 2004). Nevertheless, several studies have investigated the presence of NoV and other enteric viruses in this water type (Barthe et al. 2007; Cheong et al. 2009; Gabrieli et al. 2009; Jung et al. 2011; Lee et al. 2011; Locas et al. 2008). Although some studies did not find NoV in well water, (Cheong et al. 2009; Locas et al. 2008), most studies did demonstrate the NoV presence in various ground water sources. A study investigating Italian ground water sources found 15 % of sources positive (Gabrieli et al. 2009), while 18–42 % of samples from Korean ground water wells tested positive for NoV (Jung et al. 2011; Lee et al. 2011). Similar to the NoV screenings in river water, a broad range of NoV genotypes has been found in ground water screenings (Gabrieli et al. 2009; Lee et al. 2011). Seasonality of the NoV presence and NoV levels in ground water is not a thoroughly investigated topic, although Westrell et al. (2006) demonstrated a clear absence and the presence of NoV in summer and winter months, respectively. On the other hand, another long-term study did not find such an explicit seasonality (Lee et al. 2011). Further research is, therefore, needed to see whether NoVs display a similar seasonal pattern as enteroviruses, rotaviruses, and hepatitis A virus in ground water (Abbaszadegan et al. 2003). Regarding the use of fecal indicator organisms, no significant correlation could be shown in any study investigating the NoV presence in ground water. While some studies found the NoV presence despite the absence of indicator organisms (Barthe et al. 2007; Gabrieli et al. 2009), other studies did find indicator organisms such as E. coli and coliphages but they were not related to the NoV presence (Cheong et al. 2009; Jung et al. 2011). However, a recent study showed that a combination of different fecal indicators may be helpful, but that further research is needed to support this observation (Lee et al. 2011).

Several NoV outbreaks have been related to contaminated ground water. NoV GI.3 has been found in fecal samples and in ground water related to a NoV water borne outbreak affecting at least 84 people in Wyoming, USA (Parshionikar et al. 2003). In a water borne gastroenteritis outbreak involving bacterial and viral pathogens, 21 % of all fecal samples tested NoV positive, although ground water as cause of the outbreak could only be determined by epidemiological data (Gallay et al. 2006).

Surface Water

Surface waters (e.g., lakes, bays, and recreational water) are a known transmission route for NoV, as an overview of water borne outbreaks between 2001 and 2004 associated with surface water showed that NoV was the responsible pathogen in 39.2 % of outbreaks with a confirmed etiology (Dziuban et al. 2006; Yoder et al. 2004). Overall, only a limited number of studies have investigated the NoV presence in surface water. A Finnish study found 1 out of 7 investigated lakes positive for NoV between 2000 and 2001, while 50 % of samples from an estuarine bay in Singapore tested positive for NoV (Aw et al. 2009; Horman et al. 2004). Similar to drinking water and ground water, fecal and chemical indicators have not been related to NoV contamination of surface water (Horman et al. 2004).

Zoonotic Transmission

The existence of a zoonotic reservoir for human infective NoV has been investigated, but zoonotic transmission of human infective NoV has thus far never been proven (Li 2012). Nevertheless, the detection of human infective NoV in animal species in close contact with human beings raises suspicion for cross-species or zoonotic transmission and for the existence of an animal reservoir for human infective NoV (Mattison et al. 2007; Scipioni et al. 2008). Therefore, the animal and human infective NoV presence studies have been performed in different animals and an overview of these studies, based on the molecular detection of NoVs in feces from cattle and pigs, is presented in Tables 2 and 3, respectively. As the current review involved cooperation of a number of Belgian laboratories investigating animal NoV, we were able to compare Belgian data to worldwide observations. For the Belgian perspective, diarrheic fecal samples from domestic animals were screened for the NoV presence between 2007 and 2008. NoV sequences were detected in various animal types. While animal NoV were found in cattle and pigs, all other domestic animal samples (poultry, sheep, equine, cats, and dogs) tested negative (Axel Mauroy, personal communication).

Table 3 Overview of porcine NoVs detected worldwide

In Belgium, 7.5 % (in 2007) and 9.3 % (in 2008) of feces samples coming from diarrheic calves and young stock tested positive for the presence of bovine NoVs, with NoV GIII.2 being by far the most prevalent genotype in both studies (Mauroy et al. 2009a, b).

This data were in accordance with previously published studies from the United Kingdom, Hungary, Germany, and South Korea (Table 2) although other epidemiological data available from calf farms in industrialized countries indicated NoV prevalence ranging from 1.6 to 72 %. Similar to the Belgian results, a majority of NoV GIII.2 was also observed in most studies which may support the idea of the existence of predominating NoV genotypes that dispose of advantages upon other genotypes similar to what has been observed for GII.4 NoVs in humans (Kroneman et al. 2008; Siebenga et al. 2007; Verhoef et al. 2010, 2009). Interestingly, a single Belgian cattle sample showed a GIII.1/GIII.2 co-infection while a potential recombinant strain between NoV GIII.1 and NoV GIII.2 was found in another sample. This recombinant strain was closely related to a previously described strain NoV GIII Thirsk in British and Norwegian cattle (Jor et al. 2010; Oliver et al. 2004).

A 3 month survey found 4.6 % of Belgian pigs positive for porcine NoVs in 2007 (Mauroy et al. 2008). Although studies conducted in other industrialized countries found the NoV presence in pigs ranging between 0.2 and 25 %, Belgian results are congruent with data from Hungarian, Canadian, and Japanese pigs. Both NoV strains detected in Belgium clustered within the NoV GII.19 genotype. To date, porcine NoVs cluster into three GII genotypes and all three genotypes have been detected in Europe, America, and Asia (Wang et al. 2005b).

Worldwide and in Belgium, bovine and porcine NoV are widely endemic in their respective hosts, although data diverges greatly between studies (Tables 2, 3). These disparities observed could be explained by the use of different detection methods, the use of internal amplification controls for the detection of false negative results and different sampling strategies.

The clinical impact of bovine NoVs included watery diarrhea (Jor et al. 2010; Otto et al. 2011) while porcine NoV often causes an asymptomatic infection (Scipioni et al. 2008). Porcine NoV has exclusively been detected in fecal samples of finisher pigs (>10 weeks old pigs) (Scipioni et al. 2008; Sugieda et al. 1998; van der Poel et al. 2000; Wang et al. 2005a). Finally, murine NoVs cause a mild gastroenteritis in mice (Liu et al. 2009a; Thackray et al. 2007). Noteworthy, multiple studies have already carefully suggested that recombination of porcine and human infective NoV, both belonging to genogroup II of the Norovirus genus, could be a hypothetical zoonotic transmission route (Bull et al. 2007; Phan et al. 2007; Wang et al. 2005a). Similarly, recombination between canine and human infective GIV NoVs may be possible (Martella et al. 2009), but more data is needed to confirm whether this could lead to zoonotic transmission of NoV. Interestingly, recombination on the ORF1/ORF2 junction has also been observed among murine NoVs (Matthijs et al. 2010).

Regarding the infection of animals with human infective NoV, several studies have demonstrated that this is possible in non-human primates, pigs, and pet dogs, resulting in no clinical symptoms to very mild gastroenteritis (Bok et al. 2011; Rockx et al. 2005; Summa et al. 2012b; Takanashi et al. 2011). On the other hand, antibodies for bovine NoVs have been detected in veterinarians and in the general population (Widdowson et al. 2005a). However, the number of studies investigating cross infection of animal and human infective NoV are limited and further research is needed to see whether human infective NoV can indeed be spread via animals.

Conclusions

Development of methods for molecular detection of human and animal infective NoVs has confirmed NoV transmission routes that were suspected by investigating NoV outbreaks before sequencing of the NoV genome (Greenberg et al. 1979; Gunn et al. 1982; Koopman et al. 1982; Sekla et al. 1989; White et al. 1986). Person-to-person transmission of NoVs is by far the dominant transmission route, but food and water borne transmission are also frequently observed transmission routes. A strategy to reduce or even prevent person-to-person transmission (and NoV transmission in general) could be the development of a NoV vaccine, based on the use of virus-like particles containing capsid antigens. Some hurdles such as long-term immunity and the need for immunity against multiple NoV genotypes exist, but comprehensive strain surveillance e.g., via web-based open-access genotyping tools (Kroneman et al. 2011) such as Noronet (http://www.noronet.nl/noronet) and the use of multivalent vaccines may be a suitable approach for a vaccine development (Atmar et al. 2011; LoBue et al. 2006; Vinje 2010). Transmission of NoV by food handlers can easily lead to a NoV gastroenteritis outbreak. Both the Codex Alimentarius (2003) and the Food Code (2009) state that food handlers suffering from NoV gastroenteritis should not be allowed on the food preparation and handling areas. However, transmission of NoV via asymptomatic infected food handlers or via food handlers recovering from NoV gastroenteritis could be severely reduced by applying and respecting intervention measures (Mokhtari and Jaykus 2009). Regarding an approach to prevent NoV food borne outbreaks caused by pre-harvest contamination of bivalve molluscan shellfish, a critical limit has been suggested (Lowther et al. 2012), although both economical and public health aspects should be kept in mind (Anonymous 2012). The use of sensitive molecular NoV detection methods has shown that a substantial fraction of fresh produce can be contaminated with NoV while no associated NoV illness was reported (Baert et al. 2011). A possible solution could be confirmation of real-time PCR results using a different molecular detection assay (e.g., different primers and probes) (Baert et al. 2011). Furthermore, NoV can effectively be spread via water and has resulted in several NoV water borne outbreaks (Maunula 2006). As most water treatment plants are designed for removal of bacterial pathogens, NoV can still be present in treated wastewater (Katayama et al. 2008; Laverick et al. 2004; Rodriguez-Lazaro et al. 2011; van den Berg et al. 2005). However, further research is needed to determine whether novel methods can sufficiently reduce the NoV presence in different water types (Springthorpe and Sattar 2007). Regarding the role of animal NoVs, molecular detection coupled with DNA sequencing has allowed numerous molecular epidemiological studies of NoVs in humans and in animals (Tables 2, 3). These studies have led to the conclusion that NoV zoonotic transmission is unlikely to happen although it cannot be excluded.