Human adenoviruses associated with respiratory illness in neonates, infants, and children in the Sousse area of Tunisia

Abstract Background The human Adenovirus (HAdV) is a common agent of acute respiratory infections (ARIs). Its clinical impact in immunocompetent children and in the context of coinfections remains unclear in Tunisia. Material and methods HAdV‐ARIs were studied in hospitalized patients from birth to the age of 5 years from 2013 to 2014. Clinical and demographic characteristics, coinfections, and molecular characterization of HAdV were established. Results HAdV‐positivity was detected in 114/583 specimens (19.6%) including 6.1% single infections and 93.9% coinfections. Adenoviral coinfections mostly comprised human Rhinovirus (50.9%), Streptococcus pneumoniae (34.2%), human Respiratory Syncytial virus A/B (29.8%), and human Coronaviruses (21.9%). HAdV infection was predominant in the pediatric population (25.0% vs 10.0% in neonates, P < .001) and peaked in February 2014 (21.1%). HAdV severity of pediatric cases is characterized by low saturation of oxygen (<94%, 33.8%, P = .05) and long duration of oxygen support (≥5 days, 32.7%, P = .02). Severe HAdV infections were described with S. pneumoniae coinfection, which seemed to increase the risk of death. HAdV genotyping identified HAdV‐C as the most common species. Severe ARIs were observed in all HAdV‐identified types. Phylogenetic analysis revealed that sequences were variable suggesting the circulation of different HAdV strains sharing more similarities to strains circulating in Europe or Asia than those from Africa. Conclusion This first molecular study of HAdV in Tunisia demonstrated that it has an important role in severe ARIs with HAdV‐C being the most common species. S. pneumoniae codetection seems to increase the severity of HAdV‐ARIs.


| INTRODUCTION
Acute respiratory infections (ARIs) are a significant cause of childhood mortality and morbidity worldwide. 1 Various respiratory pathogens are identified as causative agents for respiratory symptoms including the human Adenovirus (HAdV). This pathogen is classified among the Adenoviridae family and belongs to the genus Mastadenovirus. It is a ubiquitous nonenveloped virus of medium-sized double-stranded-DNA ranging from 34 kb to more than 37 kb, which encodes around 40 genes. 2 HAdV has been divided into seven species (A-G) based on a biological criteria and DNA homology. Each species comprises multiple genotypes.
According to the Human Adenovirus Working Group, the collaboration between adenoviral researchers and the National Center for Biotechnology Information (NIH)/GenBank, up to 90 genotypes of HAdV have been described. These genotypes were designated with a number according to the chronological order of identification. 3 The distribution of HAdV genotypes is variable, depending on geographical, environmental, and meteorological characteristics. Some strains may have a higher epidemic potential. 4 Several species were found to be associated with different clinical profiles. Those most common genotypes implicated in ARIs belong to HAdV-C (types 1, 2, 5, and 6), HAdV-B (types 3 and 7), and HAdV-E (a single type 4) species. Severe and fatal adenoviral diseases were found to be caused by HAdV-B types 14, 21, and 55. 5 However, HAdV circulating genotypes in the context of ARIs have not yet been described in Tunisia.
HAdV was found to be one of the causative agents of tonsillopharyngitis, conjunctivitis, pneumonia, gastroenteritis, hepatitis, and hemorrhagic cystitis. In terms of ARIs, this virus causes between 5% and 15% of the overall respiratory illnesses in children aged below 4 years. 6 Severe or life-threatening respiratory diseases in immunocompromised cases were frequently reported in the context of adenoviral infection. In addition, HAdV severe infection was also described in immunocompetent patients. 7 Nonetheless, its detection in respiratory specimens does not necessarily identify HAdV as the causative agent of ARIs but may be related to reactivation or latency, especially, when highly sensitive molecular detection tests are used. 8 Conversely, some authors have found that there is enhanced severity and worse outcomes in children with mixed respiratory viral infection as compared with those with single infections. 9 Thus, the role of adenoviral coinfections in ARIs remains controversial and warrants additional research.
The detection of HAdV and the identification of its pathogenicity in ARIs were poorly reported in North Africa. [10][11][12][13] Although in Tunisia HAdV-ARIs were previously described, [14][15][16][17] the molecular characterization of this pathogen was only performed in non-ARIs. [18][19][20][21] This study aimed to assess the severity of HAdV and detected coinfections in neonates, infants, and children hospitalized for ARIs in a Central-East region of Tunisia. The molecular characterization of HAdV was determined. This study highlights the importance of HAdV detection and its coinfections in ARIs and provides the first starting point on the circulation of different genotypes in Tunisia. Such evidence will promote the initiation of preventive measurements against the circulation of HAdV and the most prevalent coinfected pathogens in the community.

| Ethics and study population
This study did not involve any human experimentation. The study protocol was approved with formal authorization from the Scientific and Ethical    Table 1 shows the primer sets used for HAdV molecular identification. Typing reactions were assessed starting from a first PCR (1st PCR) using universal primers designated according to the hypervariable region (HVR) of the hexon gene for all HAdV prototypes (primers set 1). A nested PCR (1st nested PCR) for the typing of HAdV-C2; -B3; -E4; -B7; and -B11 was carried out (primers sets 2-6).

| Sequencing of HAdV positive specimens
PCR for sequencing a partial region of the hexon gene specific to all HAdVs prototypes and containing the HVR-7, was performed according to Sarantis et al. 24 In addition to Tunisian samples, two specimens from Innsbruck were used as control for the optimization of HAdV sequencing reaction (see Table, Supplemental Digital Content 1 that summarizes the HAdV reference sequences targeting a partial hexon gene and control specimens used to generate the phylogenetic tree). Each specimen was analyzed based on Sanger sequencing technology using AD1 and AD2 primers, 24 accordingly. (95% CI) were calculated. The binary logistic regression model was applied when the test exceeded two tails. A value of P ≤ .05 is considered statistically significant. Continuous variables were represented using the mean ± SD or median interquartile range. The categorical variables were described using percentages. A subgroup analysis was carried out to determine P-values between the groups by pairwise comparison of the subgroups. The statistical calculations between HAdV coinfections and clinical data/severity of respiratory infection were uniquely considered for an effect size equal to or above 25 based on the minimum relevant effect size. 28

| Detection of HAdV and association with clinical and severity parameters
Respiratory pathogens were found in 526/583 of specimens (90.2%), from which 28.5% were caused by a single pathogen and 71.5% by at least two pathogens. HAdV was found in 114 patients (19.6%) and was the fourth most prevalent respiratory pathogen detected after HRV with HAdV infection, a low saturation of oxygen (<94%) was recorded with no statistical association. However, a statistically significant association was found between HAdV infection and long duration of oxygen support (≥5 days), (32.7%, P = .02). Although hospitalized due to ARIs, HAdV infected neonates had, in addition, digestive symptoms (P = .02), and HAdV infection in pediatric patients was found to be statistically associated with gastroenteritis (P < .01) ( Table 2).

| Predictive adenoviral loads and impact on clinical and severity parameters
The association of clinical data and severity of ARIs with adenoviral loads (mean and median HAdV-Ct-values) showed no statistical significance, except for anemia, which increased significantly with higher adenoviral loads (see Table,  and HCoV group and are given in Table 3. HAdV-S. pneumoniae was the only combination found to be associated with death (five fatal cases with HAdV-S. pneumoniae positive coinfection vs two fatal cases positive for HAdV (one monoinfection and one case together with another virus, but negative for S.pneumoniae). Due to the small sample size, no statistical calculations were performed.  .02  6,29 This study has discovered that HAdV seems to increase the risk of low oxygen saturation and the duration of oxygen support in the pediatric environment which is in line with other reports

| Type identification of HAdVs and phylogenetic analysis
showing that HAdV is associated with severe clinical manifestations. 30,31 In Africa, HAdV was associated with influenza-like illnesses in several countries like Morocco, Egypt, Cameroon, and Senegal. 13,[32][33][34] However, the severity of respiratory infections was variable. In Cameroon, HAdV was not found to be associated with severity. 33 Conversely, in Senegal, HAdV infection was statistically correlated with myalgia, cough, diarrhea, headache, rhinitis, and pharyngitis. 34 In the present study, a statistically significant association was found between showed no association, suggesting that viral superinfection had no additional pathogenic role, supporting other findings. 37 On the contrary to viral coinfections detected in this study, more severe infections were described in combination with HAdV-S. pneumoniae, which seems to increase the risk of death. In corroboration to our finding, it was demonstrated that HAdV, particularly types 1, 2, 3, and 5 enhance pneumococcal adherence to the human respiratory epithelial cell culture due to an upregulation of receptors for S. pneumoniae. 38 .76 Admission to ICU 4 (7.1) 6 (10.3) .  Most of the HAdV-positive infections detected in this study were HAdV-C followed by HAdV-B. A report from Cameroon found that HAdV-B infections were more common and severe than HAdV-C. 33 On the contrary, studies conducted in Egypt and Senegal were consistent with our findings showing a predominance of HAdV-C. 10,34 The most frequently identified types in this study were HAdV-C1, -C2, and -C5.
This data is in line with the epidemiology of HAdV worldwide, which found these types in children with ARIs. 45 The HAdV-C infected patients identified in the present study had severe clinical manifestations with one fatal case infected by HadV-C5. A study in the United States determining the risk factors for severe HAdV infection found that HAdV-C5 increased the risk of severe disease. 46 In addition, HAdV-B species including types 3 and 7, and in some cases types 14, 21, and 55, and HAdV-E species with the single known type 4, were also associated with severe clinical forms in the pediatric population requiring hospitalization. 36,47,48 Sequences identified in the present study were variable suggesting the circulation of different HAdV strains. Unfortunately, comparison with previous HAdV strains identified in ARIs in Tunisia was not possible as no Tunisian HAdV sequences were published. Here, only sequences concerning enteric or environmental HAdV were available. 19,20,49,50 Although the multiplex qRT-PCR can detect simultaneously a wide range of respiratory pathogens, low viral loads are probably not indicative of viral pathogenesis. The standard PCR used for typing and sequencing is based on the hexon gene, which allows more specific differentiation of adenoviral subtypes. However, it has the disadvantage of low sensitivity and requires samples with high viral loads. In addition, the fewer number of HAdV genotypes and sequences covered by the present study are considered among the study limitations. This number should be increased by referring to cell culture for samples with low viral loads or/and increasing the sampling size of the study population.

| CONCLUSION
This study provides the first molecular and epidemiological descrip-