Prevalence and Genetic Diversity of Enteric Viruses in Children with Diarrhea in Ouagadougou, Burkina Faso

Enteric viruses are a major cause of diarrhea in children, especially those under five years old. Identifying the viral agents is critical to the development of effective preventive measures. This study aimed to determine the prevalence and genetic diversity of common enteric viruses in children under five years old in Burkina Faso. Stool samples from children with (n = 263) and without (n = 50) diarrhea disorders were collected in Ouagadougou, Burkina Faso from November 2011 to September 2012. Rotavirus, norovirus, sapovirus, astrovirus, adenovirus and Aichivirus A were detected using real-time or end-point (RT-)PCR. Rotavirus strains were G and P genotyped by multiplex RT-PCR and other viral strains were characterized by sequencing of viral subgenomic segements. At least one viral agent was detected in 85.6% and 72% of the symptomatic and asymptomatic patients, respectively. Rotavirus (63.5%), adenovirus (31.2%) and genogroup II norovirus (18.2%) were the most prevalent viruses in symptomatic patients, but only rotavirus and genogroup II norovirus were significantly associated with diarrhea (OR: 7.9, 95%CI: 3.7–17; OR: 3.5, 95%CI: 1–11.7, respectively). Sapovirus (10.3%), astrovirus (4.9%), genogroup I norovirus (2.7%) and Aichivirus A (0.8%) were less prevalent. The predominant genotype of rotavirus was G9P[8] (36.5%), and the predominant norovirus strain was GII.4 variant 2012 (71.4%). Among sapovirus, the genogroup II (87.5%) predominated. Astrovirus type 1 (41.7%) was the most frequent astrovirus identified. Aichivirus A belonged to the three genotypes (A, B and C). Enteric adenoviruses type 40 and 41 were identified in 10.2% and 5.1% respectively. Several cases of co-infections were detected. The results highlight the high prevalence and the high diversity of enteric viruses in Burkinabe children.


Introduction
The study protocol and consent procedure was approved by the Ethics Committee of Burkina Faso and the ministry of health.

Sample collection and viral RNA/DNA extraction
This study was based on a relative low socio-economic status urban population of Ouagadougou (capital of Burkina Faso) consulting at Peripheral Health Centers.
Between November 2011 and September 2012, stool specimens were collected from 263 children under 5 years of age consulting for treatment of gastroenteritis in three hospitals in Ouagadougou: District hospital of Bogodogo (61 children), pediatric clinic "les Tissérins" (128 children) and Medical center with surgical antenna "Paul VI" (74 children). Diarrhea was defined according to the WHO criteria for children as the occurrence of three or more loose, liquid, or watery stools within a 24-hour period. Control stool samples were collected from 50 randomly selected children under five years old coming to the same health centers at the same period for routine immunization and not presenting gastroenteritis/diarrhea symptoms. One stool sample was collected from each patient and stored at -20°C until analysis. Viral nucleic acids were extracted from 20% fecal suspensions in phosphate-buffered saline using the Nucli-SENS 1 EasyMAG TM platform (bioMérieux, Marcy l'Etoile, France) according to the manufacturer's instructions.

Genotyping of viral strains
All positive samples by real-time (RT-)PCR were characterized by end-point (RT-)PCR. RVA positive samples were G and P genotyped using multiplex RT-PCR according to the EuroRota-Net methods (www.eurorota.net/docs.php). Caliciviruses were genotyped using primer sets SR80/NVP110 [29] and JV12/JV13 [30] to amplify a fragment of the RNA polymerase genes of SaV and NoV, respectively. Primer sets G1SKF/G1SKR and G2SKF/G2SKR [31] were used to amplify a fragment of the capsid genes of NoV-GI and NoV-GII, respectively. AstV positive samples were typed by sequencing Mon244 and Mon245 [32] RT-PCR products. All RT-PCRs were performed using the Qiagen OneStep RT-PCR kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. AdV positive samples were typed by PCR using the Ampli-Taq 1 DNA Polymerasewith Buffer II (Applied Biosystems, Foster City, CA, USA) and the primer set Hex1DEG/Hex2DEG followed by nested PCR with primers Hex3DEG/Hex4DEG [33].
Genotyping of NoV (GI and GII), SaV, AstV, AdV and AiV was performed by direct sequencing of the PCR products with the same primers used for amplification by using an ABI PRISM 1 BigDye 1 Terminator Cycle Sequencing Kit on a 3130XL DNA Genetic Analyzer (Applied Biosystems, Foster City, CA, USA). Sequencing of the PCR amplicons for the VP7 gene RVA of unknown G types and VP4 gene RVA of unknown P types was conducted with primers VP7-F/VP7-R and VP4-F/VP4-R respectively [34].

Sequence analysis
Viral sequences were compared with reference strains available in the European Nucleotide Archive (ENA) database using the FASTA program from the European Bioinformatics Institute (www.ebi.ac.uk). For norovirus, genotyping was also determined using the Norovirus Genotyping Tool version 1.0 [35].
Phylogenetic analyses were performed with the Molecular Evolutionary Genetics Analysis (MEGA) software version 6.0 software packages [36]. Phylogenetic trees were constructed by the Maximum Likelihood method with the Kimura 2-parameter as substitution model, and including the reference sequences available on the GenBank. The confidence values of the internal nodes were calculated by performing 1000 bootstrap replicates. Phylogenetic analysis of the amino-acids and nucleotides alignments was performed by using pairwise distance methods The nucleotide sequences of amplicons of Burkinabe strains were submitted to the ENA database under the following accession numbers: RVA G6 and G12 detected (VP7 gene, 842 bp, LN612503 to LN612530), NoV-GI (ORF 2, 302 bp, LN612531 to LN612535 and ORF 1, 285 bp, LN612550 to LN612552), NoV-GII (ORF 2, 302 bp LN612536 to LN612549 and ORF 1, 285 bp LN612553 to LN612571), SaV (ORF 1, 277 bp, LN612572 to LN612579), AstV (ORF

Statistical analysis
Statistical analyses were performed with Stata 1 software (StataCorp release 10, 2007; College Station, TX, USA). We compared categorical data by the Chi-square and Fischer exact tests when the population was less than 5. The level of confidence was 95% for all confidence intervals (CI). The comparison of detection of viruses between the group of children with diarrhea and the control group was evaluated as an odds ratio (OR). A p-value less than 0.05 were considered statistically significant.

Detection of enteric viruses in stools
The mean age of children less than five years old with symptoms of diarrhea was 14.1 months (versus 11.2 months in control group) and the median age was 10 months (versus 6 months in control group). The sex ratio (male: female) was 1.1 for the children with diarrhea versus 0.7 for the children without diarrhea. Among the 263 stool specimens from patients with diarrhea symptoms, 225 (85.6%) contained at least one virus, versus 36 (72%) in the control group. Table 2 summarizes the number of viruses found in children with diarrhea and children without diarrhea. RVA was the most prevalent virus in symptomatic children (63.5%) and has the strongest association with disease [p<0.001; OR: 7.9, 95%CI: 3.7-17]. AdV was the second most prevalent virus in symptomatic children (31.2%), but the presence of this virus was statistically higher in the control group than in the symptomatic patients (p = 0.01). With a prevalence of 20.9%, NoV was the third most prevalent virus in symptomatic children. Among NoV, NoV-GII was the most frequently detected and was associated with disease (p = 0.04, OR = 3.5). NoV-GI, SaV, AstV and AiV circulated at lower frequencies and their prevalence in the symptomatic group was not significantly different than in the control group (p = 0.5, p = 0.4, p = 0.4 and p = 0.09, respectively).
The distribution of positive samples in children with diarrhea disorders according to age, gender and hospitalization status is shown in Table 3. No significant difference was observed according to age, gender and hospitalization status (p>0.05, data not shown). Enteric viruses were detected throughout the study period in samples from children suffering for diarrhea with a peak of infection observed for RVA, AdV and NoV during the cold season (November to February) (data not shown).
Mixed infections were found in 94 cases (35.7%) from children suffering for diarrhea (Table 4) and only two cases (4%) in control group ((AdV+AstV+ SaV and AdV+AiV). The most common mixed infections in the group with diarrhea were RVA combined with one other virus, notably AdV (45.7%) or NoV GII (18.1%). Of note, a maximum of four different viruses were simultaneously detected in stool samples from two patients with diarrhea.

Molecular epidemiology of viral infections
A relatively large proportion of "untyped sample" was found in this study. All the "untyped samples" corresponded to samples that failed to be amplified by genotyping end-point (RT-) PCR. This was due to low viral load. As seen in Table 5, Ct-values were significantly higher for the viruses that could not be genotyped, as compared to viruses that could be genotyped.

Discussion
Diarrheal disease is a major cause of worldwide mortality in children. The importance of viruses in children gastroenteritis has been underlined in several reports worldwide and the incidence ranged between 45 and 60% [12][13][14]. The present study describes detection and characterization of viruses associated with acute diarrhea in Burkinabe children. In our study at least one virus was identified in 85.6% of symptomatic cases, against 72% in the control group. Lower prevalence rates of 35% to 40% have been reported in symptomatic patients in Europe and 60.9% in Gabon [14,37]. The systematic detection of the five viruses in the current study allowed us to observe a relatively high proportion of co-infection (35.7%) in children with acute diarrhea, which is higher than previously reported in China (5.2%) [38], in America (9%) [39] and in Tunisia (15.2%) [12]. The conditions of the studies, such as the season of sampling, the socioeconomic level of the population, sampling methods and methods of viruses detection, can explain these differences in co-infections rates.
In this study, RVA was the most common virus detected and had the strongest association with disease. This finding was reported in several studies and RVA was mostly associated in severe diarrhea cases and hospitalized young children [37,40]. The prevalence of RVA in the present study is higher than in a previous study conducted in Burkina Faso, where RVA was detected in 33.8% of the patients [16]. This could be explained by the detection method used in the present study. Indeed, real-time RT-PCR has been demonstrated to be more sensitive than antigen detection by immunochromatographic test [41]. The circulating G and P genotypes showed a diverse pattern, with G9P[8] strains dominating. Previous RVA surveillance studies in Burkina Faso revealed that G1P [8] was the predominant strain from November 2008 to February 2010 [17]. Although being regionally-restricted, these results suggest that the predominant genotype of RVA in Burkina Faso can change rapidly within a short period of time.
Others studies have shown in Europe and Asia that large fluctuations in the genotype distribution of human rotaviruses occur continuously from 1 year to another or from one place to Enteric Viruses Associated with Children Diarrhea another [42,43]. G9 was detected mostly in combination with P [8], as has been reported elsewhere in the West African sub-region [44,45]. The unusual G/P type G6P [6] (16.7%), G9P [6] (10.4%) and G12P[8] (6.3%) were also found. Recently, the emergence of G6P [6] in Burkina Faso has been reported [19]. The G6 found in this study related 100% nt (100% aa) identity with RVA/Human-wt/BFA/265-BF/2010/G6P6 (JN116531) previously detected in Burkina  Faso and clustered with AF532202/ B1711. These results showed that the same G6 strains continue to circulate in Burkina Faso area. G6 are the most prevalent genotype in cattle worldwide [46], but they are rarely found in humans. However, studies in Gabon and Democratic Republic of Congo showed a high prevalence of genotype G6P [6] among children presenting with diarrhea [14,47]. Because bovine G6 strains are highly prevalent in cattle, it is possible that the VP7 gene found in this study is derived from a previous transmission event between cattle and humans. G9P [6] rotavirus strains are widespread and have been found in many countries such as Bangladesh, Ukraine, Vietnam and many authors suggested that G9P [6] is a human-animal reassortant [48][49][50]. Indeed, in the peripheral areas of Ouagadougou, as well as Burkina Faso in general, animals and humans live in close proximity, thus increasing the possibility of RVA transmission between animals and humans. This is the first report of RVA G12P [8] strains in Burkina Faso. These strains have been shown to be emerging in West Africa [51][52] and other continents [53]. G12 strains began to emerge globally, predominantly in combination with either P [6] or P [8] in several countries [53]. The emergence of G12P [8] rotavirus confirms that these strains have the potential to become a sixth common genotype across the world [53,54].
During this study, 30.7% of RVA was G and P untypable. A low viral load may partially explain this observation. Indeed, a large part of the G and P untypable samples had cycle of quantification values >35 suggesting low viral load. Considering the better sensitivity of real- time RT-PCR used for the detection compared to the end-point RT-PCR used for genotyping, RVA can be detected but not genotyped in these low viral load samples. Another possible explanation is that untypable strains have accumulated point mutations not recognized by the primers used. Finally, the presence of RVA genotypes not yet recognized could not be excluded. Great variability in circulating rotavirus has been observed in African countries from year to year [54] and many rotavirus strains from sub-Saharan Africa remain untypable.
NoV is a causative virus of acute gastroenteritis in children and has been known to contaminate food causing viral outbreaks affecting people [55]. In this study, NoV-GII was detected in 18.3% and NoV-GI in 2.7% cases of children with diarrhea. The higher prevalence of NoV GII over GI in acute gastroenteritis cases has been reported from other parts of the world [56,57]. Prevalence of NoV-GII found in this study is higher than that reported in Gabon, China and Paraguay which found 13.9%, 27.2% and 58% respectively [14,38,58].
The presence of NoV-GII was 3.5 times higher in the symptomatic group than in the control group. In our study, NoV-GII was one of the major viruses associated with diarrhea (p = 0.04) unlike NoV-GI (p = 0.5). This study supports the increasing recognition that NoV-GII is commonly responsible for sporadic cases of gastroenteritis in children [10]. Following sequence analysis, six genotypes of genogroup II were found in both symptomatic and asymptomatic patients. NoV-GII.4 was the most prevalent genotype detected during the whole study, unlike a previous study on NoV (22.2%) in another geographical region of Burkina Faso [21]. This genotype is considered as the predominant genotype responsible for the majority of outbreaks of gastroenteritis worldwide [10]. During this study, the NoV-GII.4 variant 2012 (Sydney) was frequently detected for the capsid. This GII.4 variant emerged and caused the majority of the acute gastroenteritis outbreaks worldwide [59][60][61][62]. However the NoV-GII.4 were variant 2004 (Hunter) for the RNA polymerase. This result suggests the possibility of recombination occurred in the open reading frame (ORF)1/ORF2 overlap [63].
The presence of adenovirus in the symptomatic group was significantly lower than in the control group (p = 0.01). However, in this study, adenovirus types generally associated with diarrhea were minor. Indeed, the molecular characterization of adenovirus revealed the presence of only 12 (20.3%) strains of adenovirus generally associated with diarrhea (i.e. types 40, 41 and 31) [64,65]. This finding is higher compared to a study in Republic Democratic of Congo, which found 5.5% of adenovirus (type 40/41) [66]. Others AdV genotypes found in this study are generally known causes of acute respiratory disease or keratoconjunctivitis in children [67].
There is no significant difference between the prevalence of SaV, AstV and AiV in the symptomatic and asymptomatic population. The lack of clinical history of patient was a limit of this study. However, other studies suggesting that these viruses cause asymptomatic infections or mild forms of gastroenteritis which are treated at home [68,69].
SaV was detected in 10.3% cases of children with diarrhea. This result was consistent with those of published reports that showed that its prevalence is usually much lower than norovirus [69,70]. Genogroup II SaV (87.5%) was found to be the most common genogroup, with GII.1 the most predominant strain. GII was the predominant sapovirus strain worldwide [71][72][73]. One sapovirus found in this study showed greatest sequence homologies with Sakaeo-15 (Accession number AY646855), which represents a novel GII genetic cluster [74].
AstV was found in 4.9% of children with diarrhea. This prevalence is similar to that reported in seven provinces of China [75], Germany [69] and Ivory Coast [76], which found 5.5%, 4% and 4% prevalence, respectively. In contrast, one study in Gambia and Kenya reported AstV in 9.9% of all stools collected [77]. AstV type 1 (41.7%) was the most predominant strain and Burkinabe strains presented 95.2-100% aa identity with the AstV-1 reference strain (Accession number L23513, Oxford). This human AstV type is the most common worldwide [78][79][80]. Uncommon (type 2 and 5) and rare (type 8) serotypes were also found during this study. Serotype 8 may represent an emerging AstV type worldwide [80,81].
Only four AiV A were found in this study and belonged to genotype A, B and C, suggesting that this virus is rare in children with diarrhea in Ouagadougou, Burkina Faso. Aichivirus A has been found at low incidence in patients with gastroenteritis in several regions around the world, including Europe [82] and Africa [83].

Conclusion
A limitation of this study is the small size of control group (n = 50). Furthermore, due to the large number of untypable strains found in this study, the genetic diversity of enteric viruses in Burkinabe children may have been underestimated.
Documenting the etiology of diarrhea is important to guide vaccine development, design prevention and treatment strategies. In this study, RVA and NoV-GII were the two predominant viruses associated with diarrhea. The introduction of affordable viral diagnosis in our pediatric hospitals will improve patient care by reducing the unnecessary use of antibiotics. The relative high incidence of infections reported here suggests a possible problem of hygiene, such as water contamination, or sociodemographic level of population.
Molecular characterization reported a great diversity of strains. Thus, information about strain diversity and emerging strains would also be useful for better management of preventive strategies such as vaccination. These results highlight the importance of continuous surveillance of gastroenteritis viruses in Burkina Faso.