Recombination between enteroviruses is quite frequent, and occurs typically within the same species [1, 27]. It was initially thought that the 5’UTR and P1 regions move independently during recombination, and the frequency of recombination was higher in the non-structural coding region . This lead to the conclusion that serotyping by molecular methods involving sequencing of the P1 region should be equivalent to sequencing segments of this region (VP1,2,3 or 4). Several studies have confirmed the finding that sequencing of the VP1 or VP2 regions yields similar typing results compared to sequencing the entire P1 region [1, 14, 28]. Typing using the VP4 region has yielded conflicting results including studies reporting both consistent and inconsistent results with classical serotyping [28, 29]; Perera et al. have noted that the VP4 sequence is unreliable for serotyping of viruses in species B and C . This inconsistency seems to result from recombination events involving the VP4 region . Intuitively, one may still continue to regard the segment coding for the external domains of the capsid to recombine as a block since the VP4 domain remains completely inside the capsid and is inaccessible to antibodies .
The amplicon used for serotype determination includes a segment from the 5’UTR, the VP4 region and a segment of the VP2 region; but given the possibility of recombinations outside the VP2-VP3-VP1 region, only the sequence within VP2 was used for the typing of samples in this study. It was noted that if serotype assignment is made based on Blast comparisons with the GenBank database sequences, there is a potential for inaccurate serotype assignment since contemporary isolates submitted into GenBank may have been serotyped based on the VP4 sequence, which can reassort independently of the VP2-VP3-VP1 segment. In this study, the serotype assignment was achieved by constructing a phylogenetic tree with the partial VP2 sequence obtained by sequencing the amplicon and the corresponding VP2 sequences of the original prototype strains . In spite of the genetic drift evident in the sequences of contemporary viruses, this method provided unambiguous serotype assignment. Instances of recombination within the VP2-VP3-VP1 segment have only been rarely reported. Al-Hello and colleagues reported on an isolate that was identified as echovirus 11 by genetic analysis, but could be neutralized by antisera specific for echovirus 11 or CVA-9. A peptide scan analysis confirmed the presence of epitopes recognized by both antisera, whereas a Simplot analysis failed to reveal a recombination event within the capsid . Should reports of isolates with similar properties become more frequent, the very concept of enterovirus serotype might become untenable. Notwithstanding these considerations, at this point in time it remains valuable to serotype isolates, because the serotype is determined by the neutralizing epitopes, and remains important for the molecular epidemiology of these viruses. In addition, the capsid determines the choice of the cellular receptor, consequently the serotype is an important factor to determine the pathogenicity of an isolate ; however, the remainder of the genome also contributes to pathogenicity  and complete genome sequencing is required for full characterisation of an isolate.
During the 2010 outbreak, a different serotype distribution was noted among infants (< 1year) and older individuals, this difference in distribution of serotypes between age groups has been noted before [31, 32]. The serotypes found among infants included in our study have been typically associated with neonatal disease, for example, Coxsackie B3 and B4, echovirus 9, and EV-71. It was interesting to observe one isolate with the VP2 region most similar to the newly described EV-97 . Among the rest of the population, CVA-9 was clearly the predominant serotype. Phylogenetic analysis of the VP2 region from the CVA-9 isolates showed that all the isolates had the same sequences and therefore originated from the same strain.
Comparison of the capsid regions showed the CVA-9_2003 and CVA-9_2010 isolates to be highly homologous and this was corroborated by Simplot anaysis; consequently it is speculated that the sudden increase in detection of CVA-9 in CSF samples is attributable to an increased number of individuals susceptible to CVA-9 in the population rather than to the emergence of an immune escape mutant in 2010. For an infectious agent to cause an outbreak, it is necessary that the herd immunity of the population drops below a threshold that is determined by the basic reproduction number R0. For CVA-9 to re-enter the population of Alberta in 2010, either the new isolate was an immune escape mutant against which the population had no prior immunity, or the herd immunity had declined below the threshold. Because of the very high homology in the capsid sequences between the 2003 and 2010 isolates, it does not seem likely that the 2010 isolate was an immune escape mutant.
A comparison with the sequence of the prototype CVA-9 shows that the two Alberta isolates inherited their VP2-VP3-VP1 segment from the prototype, but the VP4 segment shows sequence divergence. For all the other protein coding segments, phylogenetic trees of the nucleotide sequence were constructed to compare the Alberta strains with the prototype strains within species B, and Simplot analysis were performed. Two important observations can be made: firstly, the two Alberta isolates are highly homologous in most regions, with few changes that can be largely attributed to sequence drift; however, recombination has likely occurred in the 2C, 3A, 3C, and 3D regions. Secondly, the Alberta isolates are very different from the prototype CVA-9 in the VP4 segment (supporting the concept that VP4 can be a site for recombination). The 5’NC, structural and non-structural regions appear to be components of a mosaic genome modified from the prototype CVA-9 by recombination and drifting. Similar observations have been made on other contemporary CVA-9 strains , and indeed in other contemporary enteroviruses within species B [22, 27]. How much the pathogenicity of these CVA-9 viruses has been modified by the non-structural genes, compared to the prototype strain, is unclear .