This was a long-term monitoring study of the epidemiology and molecular characteristics of NoVs in children under 5 years of age with acute sporadic gastroenteritis in Shanghai. Herein, the total detection rate of GII norovirus-positive cases (15.4%) among outpatients was lower than that in our previous monitoring data of NoV infections in both outpatients and inpatients [20, 22]. Although the percentage of NoV infections in 2012 (25.0%) was similar to our previous data from 2006 to 2011, the annual detection rate of NoVs was characterized by a fluctuating reduction from 2012 to 2017 . This finding may be associated with the increased publicity on how to prevent NoV infection among people because of the increasing number of outbreaks of NoV infection in many areas in recent years.
From 2012 to 2017, children with GII NoV infection were mainly aged less than 2 years, which is consistent with our previous study and other studies [20, 22-25]. No signiﬁcant difference in NoV detection rates was found in girls and boys. This finding may imply that gender is not a predisposing factor for NoV infection in children less than 5 years old. In accordance with previous studies that concluded that noroviruses mainly peaked in cold seasons [20, 22, 26-28], our study demonstrated that most NoV infections were detected in autumn and winter. However, in some other areas, NoV-associated diarrhoea had a summer peak or no apparent seasonal peak, which may be connected with an increase in contaminated water and food or other unknown reasons [21, 29].
A great diversity in NoV GII genotypes was identified on the basis of the RdRp region. Among the NoV-positive cases, GII.P4 was the predominant NoV genotype in 2012, which is consistent with our previous data on outpatient children from 2010 to 2011 and western China from 2010 to 2013 [17, 20]. It is interesting to note that the prevalence of GII.P4 abruptly disappeared after 2012, while GII.Pe became the obviously dominant genotype circulating in children under five years of age with acute sporadic gastroenteritis from 2013 to 2017. This unexpected increase in GII.Pe from 2013 was also found in Huzhou, China . Based on these results, we speculated that GII.Pe, first detected in the norovirus outbreak of 2008 in Victoria, Australia, has obviously replaced GII.P4 as the leading RdRp genotype in children with sporadic gastroenteritis in Shanghai since 2013 . However, no GII.Pe was detected in Suzhou (China) and western China in 2013 [17, 32]. Close attention should be paid to the spread of this genotype in China in the future. It was surprising that GII.P12 became the second main RdRp genotype instead of GII.Pb from 2012 to 2017. No GII.Pb was detected in this study, although it was the second most predominant RdRp genotype from 2010 to 2011 in Shanghai .
Since 2014, GII.17 has become increasingly popular in several major cities in mainland China and other areas in Asia [33-36]. However, only eight GII.P17/GII.17 strains were detected from 2015 to 2017 in our study. The same detection rate of GII.P17/GII.17 in children was also reported in Huzhou and Shanghai [30, 36]. However, adults were more susceptible to GII.P17/GII.17 than children [30, 36]. The reason for this different susceptibility to infection between diverse age groups is still unclear, and further studies need to be conducted to explore the mechanism of this difference.
Similar to the NoV RdRp genotypes, diverse NoV GII genotypes were detected on the basis of the partial capsid gene. Similar to data from Korea (2013-2015), Japan (2008-2014), Chongqing and Suzhou in China (2010-2013), Lusaka Province in Zambia (2012-2013) and Vietnam (2012-2015), GII.4 was the predominant capsid genotype spreading among children from 2012 to 2017 [17, 32, 37-40]. The differences in the predominant genotypes determined by the two sequence-based typing methods indicated the importance of genotyping NoVs simultaneously by both capsid and RdRp genes, which could assist us in comprehensively understanding the epidemiology and evolution of NoVs. Although GII.4-2006b was still the predominant GII.4 variant, GII.4-Sydney_2012 became the main and only GII.4 subtype from 2013 to 2017. GII.4-Sydney_2012 was first reported in Australia, and it has been the most prevalent variant among children worldwide since 2012 . As reported in Bangladesh (2010-2014), Chongqing (2010-2013) and Jiangsu (2010-2013) in China, in Japan (2008-2012) and in Vietnam (2012-2015), GII.3 was the second most prevalent capsid genotype in our study [17, 32, 38, 40, 42].
The analysis of combined NoV GII genotypes conducted in this study demonstrated that 19 different RdRp/capsid genotypes were determined and that 12 of them were distinct in RdRp genotypes and capsid genotypes. All of the discordant RdRp/capsid genotypes were suspected to be recombinant strains, and most of them have been reported elsewhere [14, 16, 17, 23, 37, 41]. Although more analysis of the junction of ORF1 and ORF2 is needed to confirm the recombination site in our study, the results clearly suggest that this phenomenon is very common in NoVs, as observed elsewhere. Furthermore, we observed a change in the circulation pattern of the RdRp/capsid-genotyped strains, with GII.P4-2006b/GII.4-2006b being predominant in 2012, followed by an emergence and predominance of GII.Pe/GII.4-Sydney_2012 starting in 2013. GII.Pe/GII.4-Sydney_2012, first reported in Australia in 2012, has been widespread in South Africa (2012-2013), Iran (2015-2016), Botswana (2013-2015), Korea (2013) and some cities in China (2012-2015) since then [23, 31, 37, 41, 43-46]. It was surprising that, starting with 2013, GII.Pe/GII.4-Sydney_2012 had completely replaced the pandemic GII.P4-2006b/GII.4-2006b subtype, and close attention should be paid to the prevalence of this genotype.
The second most predominant RdRp/capsid genotype in 2013 and from 2015 to 2017, GII.P12/GII.3 was also reported as the main NoV genotype in Chongqing (2011-2013), China . Interestingly, this genotype was mainly observed in the Asia-Paciﬁc region, implying that the GII.P12/GII.3 pandemic may have regional characteristics [17, 47, 48]. GII.P7/GII.6, as the second most prevalent genotype in 2014, was reported in a few cases elsewhere [46, 49]. In addition, many other rare RdRp/capsid strains, such as GII.Pe/GII.3 and GII.P16/GII.2, were detected in our study. Among these strains, the GII.P16/GII.2 strain was observed only in 2017; however, this strain has become the main genotype in Japan, France, Hong Kong, and Taiwan, as well as several other cities in China during 2016 and 2017 [17, 50-54]. Consequently, it is necessary to continuously monitor those rare strains in Shanghai.
In epidemiological investigations of children of different ages, the distribution of NoV RdRp/capsid genotypes varied in the different age groups. In our study, GII.Pe/GII.4-Sydney_2012 was detected in all age groups, while other genotypes were not. Some genotypes were only detected in one age group. These results may imply that the infection of some NoV RdRp/capsid genotypes was age specific. However, more and longer surveillance of the epidemiology of NoVs in children of different ages should be conducted to illustrate this phenomenon.