Molecular characterization of rotavirus strains from children with diarrhea in Italy, 2007–2009^†

INTRODUCTION

Rotaviruses cause a large proportion of acute gastroenteritis cases in infants and young children worldwide, with a death toll of more than 500,000 yearly, mostly in developing countries [Parashar et al., 2006]. Each year, rotaviruses cause approximately 111 million episodes of acute gastroenteritis requiring home care, 25 million clinic visits, and 2 million hospitalizations, altogether constituting a major economic burden for health care systems and families in both USA and European countries [Parashar et al., 2006; Soriano-Gabarro et al., 2006; Forster et al., 2009]. These figures are consistent with the need for implementing vaccination strategies in both developing and developed countries, as recommended by the World Health Organization [WHO, 2007].

Two rotavirus vaccines have been licensed in Europe, USA, and other countries throughout the world, and have shown high efficacy in phase III clinical trials [Ruiz-Palacios et al., 2006; Vesikari et al., 2006b]. These are: (i) the live attenuated monovalent human rotavirus vaccine Rotarix (GlaxoSmithKline, London, England) constituted by a single G1P1A[8] strain [Ruiz-Palacios et al., 2006]; (ii) the live attenuated pentavalent human-bovine reassortant vaccine RotaTeq [Clark et al., 2006; Vesikari et al., 2006b] presenting human serotype G1 through G4, and P1A[8] specificities (Merck, Whitehouse Station, NJ). A remarkable decline of pediatric admissions for rotavirus diarrhea was reported in Philadelphia in 2008, and attributed to specific herd immunity consolidated by the increasing vaccine use during the previous 2 years [Clark et al., 2009]. Although this study noted that most of 2008 cases were infected with genotype G3 rotavirus, their low number allowed no inference on a possible differential efficacy of vaccination by viral serotype. Nonetheless, a high efficacy (95%) of the pentavalent vaccine towards G3P[8] infection was later demonstrated in Texas [Boom et al., 2010]. Further studies will likely generate more data, particularly in the Americas, where children immunization has been adopted on a larger scale than elsewhere [Parashar et al., 2006; de Oliveira et al., 2008].

The correlates of protection from rotavirus disease are not fully defined [Desselberger and Huppertz, 2011], but it is largely accepted by the scientific community that both outer capsid proteins of rotavirus VP7 and VP4 stimulate neutralizing antibody responses [Yuan et al., 2009], which may be important for forming complexes with the virus on the intestinal mucosa [Coffin et al., 1997; To et al., 1998] or prevent viremia and possible extra-intestinal rotavirus replication [Blutt et al., 2007].

The issue of protective immunity from rotavirus infection is related to the complexity of the virion, exhibiting three protein layers and six structural proteins, and of its genome consisting of 11 double-stranded (ds) RNA segments which encode 11 or 12 proteins [Pesavento et al., 2006]. The ability of rotavirus to exchange entire genes during multi-strain infection of cells (reassortment) is known since many years [Gombold and Ramig, 1986; Offit and Blavat, 1986], is observed during infection in vivo in humans and animals [Iturriza-Gomara et al., 2001; Watanabe et al., 2001; Maunula and Von Bonsdorff, 2002; Esona et al., 2009], and is a major cause of the exceptional genetic variability observed for clinical rotavirus strains worldwide [Santos and Hoshino, 2005]. Interspecies rotavirus reassortment may have implications for vaccine programs effectiveness because it may introduce viral serotypes typical of animals into potentially naïve human populations, independent of vaccine coverage [Gentsch et al., 2005]. The second major factor in rotavirus genetic diversity is the accumulation of point mutations generated by the error-prone RNA-dependent RNA polymerase, that can generate neutralization escape variants [Shaw et al., 1988; Ciarlet et al., 1997; Iturriza-Gomara et al., 2001; Arista et al., 2005] and misrecognition of viral genes by established genotyping primers [Santos et al., 2003; Parra and Espinola, 2006; Solberg et al., 2009].

Despite the large diversity among rotavirus strains, a broad cross-reactive immunity between viral serotypes in vitro and in vivo is supported by several studies [Gorrell and Bishop, 1999; Yuan et al., 2005; Vesikari et al., 2007], and may explain the reported efficacy of vaccines against circulating wild-type rotaviruses of G- and P-types which are not encompassed in vaccine formulations.

To track rotavirus genes evolution, zoonotic interchange and reassortment of group A rotaviruses [Matthijnssens et al., 2008a; Martinez-Laso et al., 2009; Song and Hao, 2009], a recently proposed new classification scheme considering all 11 genome segments [Matthijnssens et al., 2008a,b] has been very useful. Based on extensive genome sequencing, this system is suited for very specific studies on rotavirus phylogeny or for detailed investigations on uncommon strains emerging in the population. Multiplex RT-PCRs with diverse primer sets are normally used for characterizing fecal group A rotavirus strains into G- and P-types, relative to the genes encoding capsid (Glycosylated) VP7 and (Protease-sensitive) VP4 proteins [Gouvea et al., 1990; Gentsch et al., 1992; Iturriza-Gomara et al., 2004; Simmonds et al., 2008]. By applying this dual nomenclature method [Estes and Kapikian, 2007; Desselberger et al., 2009], at least 23 G-genotypes and 32 P-genotypes were differentiated [Trojnar et al., 2009; Ursu et al., 2009; Collins et al., 2010], including 10 G- and 11 P-types in human infections, respectively [Jayaram et al., 2004; Desselberger et al., 2009]. Molecular typing methods have improved the understanding of rotavirus strain diversity, and permitted establishment of large molecular epidemiology programs [Gentsch et al., 2005; Santos and Hoshino, 2005; Iturriza-Gomara et al., 2009, 2010], which will be valuable to control the efficacy of present vaccines towards different co-circulating wildtype rotaviruses, estimate the impact of possible vaccine-driven immune selection of wildtype strains and help shaping possible future rotavirus vaccines development [Desselberger et al., 2006].

To gather comprehensive information on co-circulating rotavirus genotypes throughout Europe, a large consortium was established in 2007 [Iturriza-Gomara et al., 2009], embracing 17 countries, using shared methods and viral databases.

The present study reports the molecular characterization of rotaviruses from hospitalized children with diarrhea during a 3-year surveillance activity in Italy, and discusses temporal and geographical differences in the incidence of infection with different genotypes.

MATERIALS AND METHODS

RESULTS

Temporal Distribution of Rotavirus Diarrhea Cases

The monthly distribution of 2,377 rotavirus patients with available date of onset is reported in Figure 1, according to Northern, Central, and Southern groups of Regions. Monthly data for the 3 study years were cumulated. Overall, most cases of rotavirus disease (85.0%) clustered between December–May, with minor variation between years (not shown). In Southern Italy, cases peaked in March–April, and in North-Central Regions the rotavirus peak was apparently broader starting in January with a tail in April (Fig. 1). Although infrequent in the summer months, rotavirus admissions were recorded throughout the year.

DISCUSSION

The Rota-Net Italy Study Group is the Italian component of the 16 country-wide EuroRotaNet [Iturriza-Gomara et al., 2009], and besides providing rotavirus genotyping data from pediatric gastroenteritis cases into the European dataset, it was established to further analyze the diversity of co-circulating rotavirus strains in Italy at a Regional level. An overview of EuroRotaNet development and its results has been described elsewhere [Iturriza-Gomara et al., 2009, 2010] indicating the increasing impact of this initiative. A parallel advancement was experienced in the interest and efficiency of Regional diagnostic partners in Italy, where the number of Regions involved expanded from 7 in 2006/2007 to 13 out of total 20 forming Italy, since 2009. The number of strains genotyped by year has increased from 451 during the first season to 1253 and 993 in 2008 and 2009, respectively.

Some variation in the number of rotavirus diarrhea cases analyzed was observed between years and Regions, due to changes in resources allocated locally in different years according to regional disease prioritization plans and in response to public health emergencies. This was probably the case for almost all collaborating partners in 2009, when the H1N1 influenza pandemic threat arose and attracted most epidemiological attention, funds for laboratory diagnostics and hospital admissions. The number of 701 samples per year recommended in the EuroRotaNet sampling scheme was not reached in 2006–2007, due to the fewer regions enrolled during the starting year of the study. Nonetheless, as many as 2,697 rotavirus cases in 3 consecutive years were collected, beyond the total sample size (2,103) designed for early detection of emerging rotavirus genotypes with a prevalence of less than 1% in Italy [Iturriza-Gomara et al., 2009].

Since the total number of acute gastroenteritis cases tested in Regional Hospitals is undefined, this study cannot supply prevalence data on rotavirus in childhood. However, other specific studies conducted in limited areas of Italy confirmed rotaviruses as the major cause of severe pediatric diarrhea, as it was in the “1980s and 1990s” [Ruggeri and Declich, 1999; Medici et al., 2006; Forster et al., 2009; Giaquinto and van Damme, 2010]. Clear epidemic waves of rotavirus acute gastroenteritis were found during late winter-spring in the whole Italy, although a few patients with rotavirus infection were detected all-year round. In the South of the country, cases seem to peak with approximately one-month delay with respect to both Northern and Central Regions. Although this difference is small, it might however reflect a delayed winter in the South. This observation seems to be in loose agreement with recent findings that the rotavirus winter epidemic wave in Europe [Iturriza-Gomara et al., 2010] starts in the South-west heading on the North-east, similar to the experience in the US [Turcios et al., 2006].

The large genotype diversity (>40 G and P combinations) observed in all European countries [Iturriza-Gomara et al., 2010] is only partially reflected in Italy, suggesting that minor rotavirus strains may circulate in individual countries only. Conversely, the present data confirmed that in Italy apparent infections with “common” genotypes (G1P[8], G2P[4], G3P[8], G4P[8], and G9P[8]) occurred at an incidence rate very similar to the whole of Europe (89.1% compared to 89.5%). The same specific rates of “common” genotypes in Italy resembled closely those found for the whole European network, being 44.9% vs. 48.4%, 8.5% vs. 10.1%, 3.1% versus 4.3%, and 10.2% versus 15.1% for G1P[8], G2P[4], G3P[8], and G4P[8] viruses, respectively. Only G9P[8] strains were considerably more prevalent in Italy, being found in 22.4% of cases compared to an average of 11.6% in Europe [Iturriza-Gomara et al., 2010]. A similarly higher circulation of G9P[8] strains (25.1%) has also been reported in the bordering France during the same period [de Rougemont et al., 2011]. Compared to the results for Italy, the French investigation showed a higher mean prevalence of infections with G1P[8], that is, 59.2%, and remarkably fewer G4P[8] cases (2.2%). However, also in the case of France, the vast majority of rotaviruses typed belonged to the common types, and no obvious emergence of human-animal reassortants or zoonotic strain was detected [de Rougemont et al., 2011].

Differences in the prevalence of circulating genotypes described among European countries [Iturriza-Gomara et al., 2010] were also observed between the different Regions of Italy, and between years. To minimize fluctuations due to the low sample size of specific Regions, data were cumulated according to Northern, Central, and Southern groups of Regions (for details of grouping of the regions, see Table I). This grouping also somehow reflects differences along a North-to-South gradient in both socio-economic (industrialized/metropolitan territory towards more rural/smaller cities) and climatic (cooler/humid towards warmer/dry) parameters. Southern Regions are also more exposed to immigration particularly from Africa and other Mediterranean countries. This might partially explain the occurrence of “untypable” rotaviruses mostly in the South (4.1–5.3%), throughout the study. Accordingly, Southern Italy showed a lower (by 5–7%) frequency of “common” genotypes, and a significant decrease in G1P[8] rotavirus prevalence throughout the study period, particularly in 2006–2007 (20.9% vs. 73.6 and 61.8% in Northern and Central Regions, respectively). This was compensated by more G2P[4] and G4P[8] strains in the South (14.5% and 22.6% vs. national means of 8.5% and 10.2%, respectively). The only peculiarity of Northern Italy was a higher circulation of G9P[8] rotavirus compared to the other Regions in 2007–2009, whereas this genotype was more frequent in the South in 2006–2007. This large variability of co-circulating rotavirus genotypes by time and space in Italy remains unexplained, as it is in other countries [Santos and Hoshino, 2005]. The possibility of a differential selective pressure of vaccines towards emergence of individual genotypes cannot be invoked at present because rotavirus vaccination has progressed very slowly in all Regions of Italy. It is possible that strains which become predominant in a particular area and year would subsequently re-emerge depending on both their residual persistence in healthy carriers, animal reservoirs or the environment and/or on the immune pressure previously established in the population. It is also possible that novel strains may be locally introduced from other areas displacing formerly predominant strains based on limited antigenic differences and/or fitness to replicate in humans. As a matter of fact, it is not clear to what extent genotype diversity truly reflects diversity in host (immune system)-pathogen relationships, and the basis of rotavirus attenuation is still controversial, although it is tempting to believe that lower replication in vivo of vaccine or animal-derived strains infecting man might be involved [Vesikari et al., 2006a; Anderson, 2008; Feng et al., 2011].

Large vaccination trials and first analyses of national campaign efficacy indicate that both commercial vaccines adopted in Europe including Italy are highly efficacious against disease caused by the major rotavirus genotypes [Ruiz-Palacios et al., 2006; Vesikari et al., 2006b, 2007, 2009; Perez-Schael et al., 2007; Phua et al., 2009; Zaman et al., 2010]. Therefore, the finding that common rotavirus genotypes represent the large majority of circulating strains in Italy suggests that mass vaccination may be expected to have a significant impact on rotavirus acute gastroenteritis also in this country, as reported elsewhere. Whether vaccine efficacy will be different in different areas of Italy as a consequence of the local genotype diversities observed in this study cannot be predicted at present.

The persistence of the diversity of genotypes observed suggests per se that the situation may change easily. Novel strains carrying uncommon G and/or P genes as well as other genomic differences may emerge, as increasing exchange of people, food and animals between continents might eventually lead to importation of strains with new antigenic and virulence characteristics and particular fitness to generate epidemic or pandemic outburst of disease. Novel rotaviruses with reassorted gene segments may arise from co-infection of human or animal hosts [Gentsch et al., 2005; Martella et al., 2010]. In fact in the present study mixed infections were found in 7.6% of rotavirus patients, that may at least partially explain the 2.0% of “uncommon” rotavirus strains detected. Half of these may be derived from reassortment of “common” G and P types [Iturriza-Gomara et al., 2001], whereas the rest of them involved gene exchange from animal rotavirus strains, particularly P[6] or G[10] as recently shown by other authors [Steyer et al., 2008, 2010; Banyai et al., 2009a,b, 2010; Mukherjee et al., 2010].

The probability and outcome of mixed infections with two different rotavirus strains may depend on both the frequency of each strain co-circulating in the population, and possible interference phenomena during dual replication in the gut [Ward et al., 1996]. To explore these events, the observed numbers and genotypes of mixed infections involving common rotaviruses in Italy during the two seasons of 2007/2008 and 2008/2009 were compared with the numbers and genotypes of detected strains likely to have arisen by natural reassortment. Different types of reassortant rotaviruses were found to be grossly proportional with the numbers of observed relevant mixed infections. However, in this assessment the geographical or temporal incidence of mixed infections or natural reassortants was not considered, and thus, although logical, the correlation of low significance (P < 0.05) found is not one which will be discovered with necessity. In this context it should be noted that Muhsen et al. [2009] recently reported data from Israel where a significant correlation between the numbers of relevant mixed rotavirus infections and of natural reassortants likely to have emerged from them has been observed.

The occurrence of 37 cases where genotyping was unsuccessful might be explained by limited virus concentration in some samples and/or minor variation in the nucleotide sequence targeted by the PCR primers. Nonetheless, it cannot be excluded that some of these viruses, otherwise confirmed as rotaviruses by amplification of the VP6 encoding gene, might carry significantly different G and/or P specificities, and represent uncommon genotypes either deriving from animals or imported from abroad. In fact, the multiplex PCR used for human rotavirus detection does not include all primers needed to cover the complete genotype spectrum of group A rotavirus affecting animals, and is optimized for human strains [Iturriza-Gomara et al., 2010]. Supplementary study will be needed to characterize these untypable rotaviruses further, possibly by designing different primers for G or P gene amplification and/or by sequence analysis of other genes [Matthijnssens et al., 2008a,b].

Complex gene evolution and reassortment is likely required before animal rotavirus strains transmitted zoonotically can adapt to efficient replication and transmission in humans [Matthijnssens et al., 2008a; Martinez-Laso et al., 2009]. This is possibly the reason why strains with an animal G or P type, such as G8, G12 or P[6] in the present study, did not appear to spread significantly following their first appearance. Rotaviruses with both G and P type of animal origin were not detected at all. For this reason, it is also hard to address the question whether humans are the host where “humanization” of animal rotavirus strains first occurs, or an intermediate animal species other than the one of origin is required for strain adaptation to humans. Some underestimation of emerging animal rotaviruses may have occurred in this study, due to either lower viral load of animal viruses compared to homologous strains during dual infection, and/or to technical issues, such as unsuitable diagnostic or genotyping primers used in normal typing routine. This is one of the reasons of the importance of international laboratory networking, such as EuroRotaNet, in facilitating implementation, optimization and sharing of state-of-the-art detection methods and construction of large virus and typing data repositories fit to identify novel rotaviruses at an early stage of expansion in the human population.

APPENDIX