Epidemiological characterizations and clinical severity of rotavirus infections
Children with RV1 were significantly younger [median age 3.2 (2.4, 3.8) vs 20.8 (14.6, 32.8) months old, (p<0.0001)], more likely to be female (56.8% vs 38.9%, P =0.039), and to have mild symptoms and overall disease severity than those with wildtype G1P[8] (Table 2). Of all genotypes, RV1 was identified most frequently in children under 3 months of age (~10% vs. ~1% for all others) (Figure 3). Children with the emergent G12P[8] genotype were older (24-59 month) and more often female (53.2% vs 38.9%, P =0.02) than those with wildtype G1P[8] (Table 2; Figure 3). However, there was no increase in the proportion of participants who were older during the 2017-2018 epidemic season when G12P[8] caused the majority of rotavirus infections (Figure S1). Children with G2P[4] were less likely to be hospitalized (4.0% vs 15.1%; P =0.01) and had lower MVS scores [14 (11, 16) vs 15 (12, 17); P =0.03] than those with G1P[8](Table S1). With the exception of RV1 infection, children with major rotavirus genotypes had similar clinical symptoms, management, and disease severity scores (Table 2 and S1). Furthermore, co-infection with other gastroenteritis viruses, enteric bacteria, or parasites in these children did not result in different clinical characteristics (Table 2).
In unvaccinated children, AGE was most commonly due to the four predominant genotypes each season: G1P[8], G2P[4], G9P[8], and G12P[8] but not RV1 (Figure 4). RV1 was the predominant strain, detected in more than 50% of children with AGE who had received 1 or 2 RV1 doses (Fisher’s exact test, P <0.001).
Discussion :
In the months preceding RV1 introduction, rotavirus G9P[8] was the predominant genotype with a high percentage of G1P[8] strains also detected. Following the implementation of a publicly funded RV1 vaccination program in Alberta with greater than 80% coverage(41), we observed a large reduction in the percentage of AGE due to rotavirus infections with the peak of activity diminishing during the 2015 to 2018 study years (Figure 2). While the circulating rotavirus genotypes remained mostly unchanged, G9P[8] was no longer the prevailing strain with the homotypic G1P[8], heterotypic G2P[4], and the emerging G12P[8] strains becoming predominant.
Consistent with previous reports, we found that strain predominance varies from year to year. However, more comprehensive multiyear data are need to ascertain whether a delayed detection peak, a shortened duration of the peak, or a shift from annual to biennial epidemics is observed locally post vaccination introduction as documented in the US (42). Controversy exists related to the long-term effect of vaccination on the circulating genotype diversity (43). Whether the shift in rotavirus genotype distribution in our study is due to natural fluctuations as part of the inherent evolution of the virus, or to the emergence of new vaccine-escape mutants due to selection pressure from RV1 remains to be clarified.
The genotype constellation of the RV1 isolates found in our cohort closely resemble the original RV1 virus. We did not identify any vaccine and wild type rotavirus reassortments in the small number (n=4) of RV1 strains analyzed. Improved sensitivity of Illumina sequencing for low viral load rotavirus samples is needed for future studies. RV1 strains were associated with very mild AGE in infants under three months and were detected in children who had received a single vaccine dose (19). We could not differentiate between symptomatic infection by RV1 and post-vaccine shedding of RV1 and incidental gastroenteritis symptoms in these children.
G1P[8] was the predominant genotype reported previously in a multisite genotype study, including Alberta, conducted under the Canadian Immunization Monitoring Program, Active (IMPACT) (unpublished data) in 2014 and 2015. In this study, we found that G9P[8] had the highest prevalence in the short eight-month surveillance season of 2014-2015, comprising 31.9% of strains. The G9P[8] genotype was first reported in the 1980s and has been widespread globally since the mid-1990s, circulating as a minor type or predominant strain, accounting for >70% of all rotavirus cases (17, 18).
Increases in G2P[4] prevalence have been reported in countries and regions with consistent RV1 vaccine utilization (44), raising the question of whether RV1 offers limited cross-protection against the heterotypic G2P[4] strains since the genetic backbone (i.e. genotype constellation) of the two (Wa-like vs DS1-like) are different. Although we detected a transient increase in G2P[4] prevalence in the second year following RV1 introduction, this must be interpreted cautiously as the heterotypic G2 is often associated with cyclic re-emergence during 10-year periods (44). As such, it is not clear if the rise in G2P[4] detection was due to vaccine selection pressure or simply reflecting the natural oscillation of this genotype. A review of G2P[4] strain evolution reported that alterations in G2P[4] distribution occur commonly in countries that use RV1, RV5 or mixed vaccination strategies as well as in countries without routine rotavirus vaccination (44). Our data, support the global data (44) that there has not been a surge of G2P[4] driven by vaccine selection pressure.
Strains of the G12 genotype were first described in infants with AGE in southeast Asia in 1987 and 1998 (45, 46). Since then, the G12P[8] genotype has been increasingly detected around the globe and gained predominance as the sixth epidemic strain during the past few years (47, 48). In Alberta, G12P[8] strains first emerged in 2012 and became predominant in 2018 according to provincial viral AGE surveillance data (unpublished data). We found that G12P[8] predominantly infected older children and was the most common strain in fully vaccinated children. This suggests that G12P[8] may have unique epitopes that evade host immune responses generated from vaccination or prior infection by another rotavirus genotype. Thus, RV1 vaccine may provide insufficient cross-protection against this emergent strain. Indeed, multiple antigenic mismatches were identified in the VP7 (G protein) of the predominant G12P[8] and the vaccine strains in the US (47). Consistent with our results, G12P[8] infection was more common among partially or fully vaccinated infants than other rotavirus genotypes. However, an overall reduction in rotavirus activity occurred in 2017-2018 when G12P[8] predominated suggesting good protection of RV1 against rotavirus AGE. Additional studies with longer surveillance periods and larger sample sizes are warranted to understand vaccine effectiveness against G12P[8]-specific rotavirus AGE.
Knowledge of the association between rotavirus genotype diversity and AGE clinical severity is limited. Previous studies have presented contradictory findings on this topic (18, 20-22). In our study, children with RV1 detected in their stool specimens had less severe AGE episodes, consistent with symptoms due to the live attenuated vaccine virus. We also found that children with G2P[4] infections were less likely to require hospitalization than those with G1P[8]. While the median MVS score for children with G2P[4] infections was in the severe range and clinical symptoms and other severity measures were comparable with those of G1P[8], the median total MVS score was one point lower than that of total rotavirus cases regardless of genotype. Our data suggest that other than RV1 related G1P[8] isolates, the clinical severity of different rotavirus genotypes did not differ significantly. Wild type rotavirus of all genotypes detected were found to cause severe AGE in young children thus emphasizes the need for rotavirus vaccines that protects against all genotypes.
A limitation of our study is that the assay utilized was unable to detect the new equine-like G3P[8] strains. Thus, future studies with updated primer sets (49) are needed to obtain a more complete picture of circulating rotavirus diversity. Moreover, as the rotavirus genotype prevalence fluctuates naturally over a period that varies between 3 and 11 years (50), a longer surveillance period that includes additional pre- and post-vaccine years is needed to fully understand the impact of rotavirus vaccination on local rotavirus diversity.
In summary, our study revealed dynamic changes in rotavirus genotype prevalence before and after RV1 introduction into Alberta’s routine immunization program: G9P[8], G1P[8], G2P[4] and G12P[8] predominated consecutively each season. This emphasizes the need for ongoing monitoring of vaccine effectiveness against all genotypes as most continue to be associated with severe AGE in children. Continued rotavirus surveillance and detailed whole genome characterization of strains could be employed to determine the changes in epidemiology and distribution of rotavirus genotypes and to assess the potential impact of vaccination selection pressure on rotavirus evolution in the post vaccination era.