Influence of Novel Norovirus GII.4 Variants on Gastroenteritis Outbreak Dynamics in Alberta and the Northern Territories, Canada between 2000 and 2008

Background Norovirus GII.4 is the predominant genotype circulating worldwide over the last decade causing 80% of all norovirus outbreaks with new GII.4 variants reported in parallel with periodic epidemic waves of norovirus outbreaks. The circulating new GII.4 variants and the epidemiology of norovirus outbreaks in Alberta, Canada have not been described. Our hypothesis is that the periodic epidemic norovirus outbreak activity in Alberta was driven by new GII.4 variants evolving by genetic drift. Methodology/Principal Findings The Alberta Provincial Public Health Laboratory performed norovirus testing using RT-PCR for suspected norovirus outbreaks in the province and the northern Territories between 2000 and 2008. At least one norovirus strain from 707 out of 1,057 (66.9%) confirmed norovirus outbreaks were successfully sequenced. Phylogenetic analysis was performed using BioNumerics and 617 (91.1%) outbreaks were characterized as caused by GII.4 with 598 assigned as novel variants including: GII.4-1996, GII.4-2002, GII.4-2004, GII.4-2006a, GII.4-2006b, GII.4-2008a and GII.4-2008b. Defining July to June of the following year as the yearly observation period, there was clear biannual pattern of low and high outbreak activity in Alberta. Within this biannual pattern, high outbreak activity followed the emergence of novel GII.4 variants. The two variants that emerged in 2006 had wider geographic distribution and resulted in higher outbreak activity compared to other variants. The outbreak settings were analyzed. Community-based group residence was the most common for both GII.4 variants and non-GII.4 variants. GII.4 variants were more commonly associated with outbreaks in acute care hospitals while outbreaks associated non-GII.4 variants were more commonly seen in school and community social events settings (p<0.01). Conclusions/Significance The emergence of new norovirus GII.4 variants resulted in an increased norovirus outbreak activity in the following season in a unique biannual pattern in Alberta over an eight year period. The association between antigenic drift of GII.4 strains and epidemic norovirus outbreak activity could be due to changes in host immunity, viral receptor binding efficiency or virulence factors in the new variants. Early detection of novel GII.4 variants provides vital information that could be used to forecast the norovirus outbreak burden, enhance public health preparedness and allocate appropriate resources for outbreak management.


Introduction
Norovirus (NoV) is recognized as the most common cause of gastroenteritis outbreaks worldwide. NoV outbreaks occur frequently in nursing homes, health care institutions, cruise ships and have also been reported in schools and prisons [1][2][3], causing considerable morbidity and mortality in these settings [4]. Noroviruses are extremely contagious with the minimal infectious dose as low as 10 viral particles [5] while the amount of virus shed by infected individuals is high (over 10 8 RNA copies per gram of stool) [6][7][8]. Transmissibility is enhanced by the stability of NoV in the environment and its resistance to disinfection [9]. Norovirus can be transmitted by contaminated food and water, person to person spread, and exposure to aerosols from vomitus [10][11][12]. Norovirus outbreak is therefore a challenge to manage and control and has considerable health and economic impact.
Norovirus is a genetically diverse group of virus belonging to the family Caliciviridae. There are currently five genogroups (GI-GV). GIII has only been found in cattle [13,14] and GV only in mice [15,16]. GI (8 genotypes) and GII (17 genotypes) contain most of the strains infecting humans [16]. The NoV GII.4 genotype has been the predominant genotype circulating in the US, Europe and Oceania over the past decade and has caused up to 80% of all NoV outbreaks, particularly those occurring in healthcare settings [17][18][19][20]. Recent studies have reported that GII.4 capsid sequences have evolved over the last 20 years and in parallel, NoV outbreaks of epidemic proportion have been observed [21]. The pandemic spread of new GII.4 variants was first recognized in the mid-1990s [22]. In 1995-1996, strain US95/96 was responsible for about 55% of NoV outbreaks in the US and 85% in the Netherlands [23]. Between 2000 and 2004, US95/96 was replaced by two new GII.4 variants, the Farmington Hill strain [3], which was associated with 80% of the US NoV outbreaks [1]. During the same period in Europe, the new GII.4 variant caused outbreaks during the winter, spring and summer [24][25][26]. In 2004, the Hunter GII.4 variant was detected in Australia, Europe, and Asia [26][27][28]. This strain was then replaced by two co-circulating GII.4 variants in the US, Europe, and Asia in early 2006 [4]. Between 2007 and 2008, two new GII.4 variants emerged and were reported by the food-borne viruses in Europe (FBVE) Network (Joukje J., personal communication). All NoV GII.4 variants are genetically linked and have been classified internationally as novel GII. 4 variants 1996, 2002, 2004, 2006a, 2006b and 2008 respectively.
The circulation of NoV GII.4 variants in Alberta and northern Territories in Canada and the epidemiology of gastroenteritis outbreaks has not been described. We examined nine years of gastrointestinal outbreak data (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008) and studied the emergence and circulation of new GII.4 variants and their association with NoV gastroenteritis outbreak activity.

Ethics Statement
Health ethics was not requested as this study was part of routine laboratory-based outbreak investigation and identifiable information of the outbreaks was not included in the analysis. Patient consents were not required as they were tested as per routine outbreak laboratory investigations and patient demographic information was not included in the analysis.

Laboratory investigation of gastroenteritis outbreaks
All suspected gastrointestinal outbreaks in the province of Alberta, Canada are identified and investigated through the provincial public health system using standardized protocols and tracking systems established since 1999. Between March 2004 and March 2009, the province was divided into nine health regions, including: two metropolitan regions each with a population .1 million and seven non-metro regions with populations ranging from 82.8 K to 183.5 K. The Alberta Provincial Public Health Laboratory (ProvLab) performed laboratory investigations on stool samples from all gastroenteritis outbreaks in Central and Northern Alberta between 1999 and February 2002 and for the whole province from March of 2002 onward. ProvLab also provided laboratory investigation support for gastroenteritis outbreaks in the northern territories in Canada (the Northwest Territories, Yukon and Nunavut) during this period. The settings of the outbreaks were recorded in ProvLab database and could be classified as one of the followings: group residence, senior residence, long term care, hospital long term care, hospital acute care, daycare, school, community gathering, conference, food establishment, cruise/ hotel, private households, community shelter/service, regionbased community and others.
If NoV was the suspected etiological agent for a gastroenteritis outbreak, stool specimens were processed for both standard bacterial culture and nucleic acid detection of NoV using conventional reverse transcript PCR between 1999 and March 2004 [29] and a multiplex real time RT-PCR (Mrt RT-PCR) assay since April 2004 [30]. At least one NoV positive specimen from each outbreak occurring between July 2000 and June 2008 was selected for genotyping by DNA sequencing.

RT-PCR assay and Sequencing for Norovirus
The Mrt RT-PCR assay used for NoV detection in the stool specimens from gastroenteritis outbreaks and the methods for nucleotide sequencing to characterize NoV genotype and variants were previously described by Pang et al [30,31]. Briefly, total nucleic acid from 200 ml of 10% stool suspension was extracted and eluted into 100 ml using Magazorb TM RNA extraction kit (CORTEX Biochem, CA, USA) by automatic extraction or easyMAG (bioMérieux) according to the manufacturer's instructions. cDNA was synthesized using random hexamer and Super-Script TM II RNase H -Reverse Transcriptase kit (Invitrogen, CA, USA). Five micro liters of cDNA was used for detection of NoV and the remaining cDNA sample was stored at minus 20uC for later characterization of NoV. cDNA samples tested positive for NoV was used for amplification with primers from region E of the capsid gene for GII and from region D for GI. If amplification failed with the region E primers for GII, then amplification was performed using the primers from region D for GII. The amplicons (GII region E = 320 bp, GII region D = 253, and GI region D = 177 bp) were purified using Qiagen purification kit (Qiagen, Chatsworth, CA, USA). Sequencing was performed using the BigDyeH Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) in the ABI PRISMH 3100-Avant Genetic Analyzer. Sequencing analysis utilized the Data Collection Software v2.0 (Applied Biosystems).

Phylogenetic Analysis
Primary sequence data was assembled and the consensus sequence from each amplicon was generated using the sequence software BioNumerics 5.1. The sequence data was blasted against the database (GenBank, NCBI) to confirm the NoV assignment before performing the phylogenetic analysis. Pairwise alignments and multiple alignments of DNA sequences were performed with the default parameters. A minimum spanning tree was constructed based on 320 nucleotides in the E region of the capsid gene using the neighbor-joining algorithm with BioNumerics software (version 5.1; Applied Maths Sint-Martens-Latem, Belgium).
Norovirus genogroup, genotypes and the GII.4 variants were assigned using the reference GenBank access numbers and nomenclature described by Zheng et al in 2006 [16], the FBVE network and CaliciNet, US (Dr. M Koopmans and Dr. J Vijin, personal communication) as shown in Table 1.

Definition of the observation period, setting and epidemic year for NoV outbreaks
The annual observation period for NoV gastroenteritis outbreaks was defined as the period beginning in July and ending in June of the following year to improve the capture of the winter seasonality of NoV outbreaks. Norovirus outbreaks always peaked between the months of November to March. This seasonal assignment allowed better analysis of the annual fluctuations of NoV outbreak burden and its relationship to strain variation and evolution. For this study, an epidemic year was defined as when the number of NoV outbreaks during the 12-month period was $2 times higher than the numbers of outbreaks in the period before and after the observation period. The observation periods with a relatively low number of NoV outbreaks were designated as quiescent years.
For the analysis of the outbreak settings, outbreaks in group residence, senior residence and long term care were grouped together as community-based group residence. Outbreaks in community gatherings, conference and food establishments were considered as similar setting with sharing of food and beverages and were classified as community social events.

Data Analysis
The difference in the number of NoV outbreaks by observation periods and the association of new variant strains and epidemic year was analyzed using binary logistic regression. Chi Square Test or Fisher Exact test as appropriate with Bonferroni correction was used to analyze the difference in NoV outbreaks for GII.4 variants and non-GII.4 variants in the following outbreak settings: community-based group residence, hospital long-term care, hospital acute care, community social events and school.      Table 2.

Clinical settings of GII.4 variants and non-GII.4 variants outbreaks
The three most common settings for outbreaks caused by GII.4 variants were community-based group residence (n = 427, 69.2%) followed by hospital long-term-care (n = 92, 14.9%), hospital acute care (n = 44, 7.1%). The three most common settings for non-GII.4 variant outbreaks were community-based group residence (n = 56, 62.2%), followed by schools (n = 8, 8.9%) then community social events (n = 7, 7.8%) ( Table 4). Outbreaks associated with GII.4 variants as compared to non-GII.4 variant were significantly more common in the settings of hospital acute care (p,0.01, Chi square test with Bonferroni correction) while outbreaks associated with non GII.4 variants were significantly more common in the settings of school and community social events (p,0.005, Fisher Exact test with Bonferroni correction). Most of the outbreaks in hospital acute

Discussion
By defining the annual observation period in our study in a unique way rather than by calendar year as in other studies [17,19,32], and access to a unique population-based dataset spanning eight years of gastrointestinal outbreaks investigated using standardized protocols allowed us to clearly observe a distinct pattern of alternating epidemic and quiescent years of NoV outbreak activity in Alberta and the northern Territories. This pattern was temporally associated with the emergence of distinctly different GII.4 variant strains. A persistent and temporal genetic drift in NoV capsid sequences was also reported worldwide [17,27,[32][33][34][35]. The reason for the frequent antigenic drift of GII.4 strains resulting in critical amino acids changes (antigenic drift) in the P2 domain of the viral capsid protein binding site [35][36][37][38] is not understood.
Lindesmith et al. studied the molecular mechanism of GII.4 NoV evolution resulting in the persistence and emergence of new strains [21]. The emergent variants appear to have a transmissibility advantage and increased virulence as documented by the high prevalence of NoV outbreaks caused by variant 2006b/a and the steady increase in the number of NoV outbreaks over the 8 years of this study. It has been hypothesized that GII.4 NoV persist by altering their ABH histo-blood group antigens (HBGAs) carbohydrate-binding targets over time, allowing for escape from host susceptibility alleles that are highly penetrant.
An alternate explanation is that evolving strains drift under immune selective pressure until mutations have accumulated to the point where a novel genetic variant phenotype becomes established and evade from pre-established host immunity [35]. A similar phylodynamic pattern of epochal evolution has been described for influenza A virus (H3N2) [39]. Short-term protective antibodies against NoV previously described usually waned after 6 months in the absence of re-exposure [40][41][42]. Some new variants persisted for a long period of time at a low level which could be explained by ongoing mutations that allowed partial but incomplete evasion of the host immune response but did not result in increased virulence or circulation in a totally immunologically naïve population that would result in activity levels of epidemic proportion. Further studies on the immune response to NoV within populations and the duration of protective immunity against prevalent genetic variants are urgently needed.
Early detection of novel GII.4 variants could be used to forecast the NoV outbreak burden and allows enhanced surveillance and early implementation of preventive measures in populations and settings where outbreaks would most likely occur. We observed a time lag of 4 to 6 months between the initial circulation of new variants and the outbreak peak in the subsequent epidemic year. This would be the available time frame for the development of future vaccine if such capacity becomes available for NoV.
We demonstrated clearly that a single novel variant of NoV GII.4 caused the majority of outbreaks in four epidemic years in this study, similar to other reports with variable periods of observation from different continents [17,27,[32][33][34][35]. However, the high prevalence of NoV outbreaks in the epidemic year 06-07 was caused by two of GII.4 variants that had emerged simultaneously (2006a and 2006b) and the total number of NoV outbreaks observed was higher than those identified in previous epidemic years. The global increase in NoV outbreaks in 2006 was linked to these two new variants [19,20]. In contrast with the data from New Zealand and Australia [20] but consistent with observations from studies reported in the US [4] Netherlands [19] and Eastern Spain [17], The 2006b variant appears to have emerged from an earlier strain that had accumulated more mutations after a long period of stasis. We hypothesize that an epochal change occurred in the 2006b variant resulting in a significant number of amino acid substitutions leading to a distinct phenotype with better host evasion capability and higher transmissibility. 2006 was the first time during our observation period that two new variants emerged during the same year. Ongoing surveillance of the circulatory pattern of these strains would be critical to our understanding of the dynamics of co-circulation of NoV variants.
As observed in all NoV outbreaks worldwide [42], most of the NoV outbreaks occurred in closed settings in Alberta such as community-based group residence and hospital long term care facilities. The non-GII.4 variant were found to be significantly associated with outbreaks in in schools and community social events as compared to GII.4 variants. The number of outbreaks per observation in these two settings were small but consistent during the study period (data not shown) making the observation unlikely to be a result of temporal bias. In a previous study, we reported a broader range of NoV genotypes in sporadic gastroenteritis in young children as compared to outbreaks [31]. This study further suggested that non-GII.4 genotypes were more common in outbreaks in the younger populations. The finding might reflect differences of age-related susceptibility to different NoV stains. The population in hospital acute care was at higher risk for NoV outbreaks cause by novel variants and could be an important target for molecular surveillance for new strains.
Our study supported the importance of ongoing surveillance of circulating NoV strains at a phylogenetic level with the observation that the emergence of new GII.4 variants is associated with increase in the NoV outbreak burden. Such surveillance data has the potential to enhance public health preparedness and allow planning for more appropriate health care resource allocation. Further studies are needed to better understand the relationship between the host response and emergence of new variant strains and the dynamics of strain circulation within populations both locally and globally.