The molecular subtyping of Human rhinovirus(HRV) in children from three Sub-Saharan African countries.

The pathogenesis of human rhinovirus (HRV) during severe respiratory disease remains undefined; thus, we aimed to explore the relationship between the HRV molecular subtyping results obtained during severe and asymptomatic childhood infections. Nasopharyngeal/oropharyngeal swabs from children (1 to 59 months of age) hospitalized with pneumonia and from age-frequency-matched controls were collected between August 2011 and August 2013.

There were no differences in the prevalences of HRV detection and HRV load between cases (21% and 3.6 log 10 copies/ml, respectively) and controls (20% and 3.5 log 10 copies/ml; P ϭ 0.334 and P ϭ 0.08, respectively). Furthermore, the prevalence of HRV identification, the species distribution, and the viral load did not differ between cases and controls at the individual sites (data not shown). However, among children Ͼ12 months of age, the cases had a significantly greater prevalence of HRV detection (21%) than the controls (16%, P ϭ 0.009), with the results driven by HRV-C detection (12% versus 7%, P ϭ 0.001) ( Table 1).
The seasonal circulation of the three HRV species over the full study period for cases and controls is detailed in Fig. 1. For both cases and controls, HRV-B appeared sporadically throughout the year with no obvious seasonality pattern. Similarly, there were no obvious patterns in the circulation of the HRV-A and HRV-C species among both cases and controls; however, they were detected throughout the year. Furthermore, no obvious pattern or clear seasonal distribution of HRV infection was observed at any of the sites (see Fig. S1 in the supplemental material). The overall prevalence for HRV ranged from 11% in August 2011 and March 2012 to 33% in August 2013 among cases and from 1% in October 2011 to 27% in January 2014 among controls during the study period.
HRV clinical and molecular subtyping among community controls. Among the HRV-positive samples from community controls, 97% (n ϭ 446) were successfully amplified, 3% (n ϭ 14) failed to amplify, and 0.5% (n ϭ 2) were of insufficient volume for serotyping. Among the amplified samples, 5% were typed as enterovirus and one as echovirus and 92% as HRV. The dominant HRV species among controls were HRV-C (45%) and HRV-A (45%), and 10% were HRV-B. There were no differences in species distribution between asymptomatic controls (HRV-A, 47%; HRV-B, 10%; HRV-C, 43%) and controls with signs and symptoms of upper respiratory tract infections (including runny nose, cough, wheeze, difficulty breathing, and fever) (HRV-A, 38%; HRV-B, 6%; HRV-C, 56%; P ϭ 0.066). There were no differences in demographic and clinical symptom characteristics among controls with HRV-B infection compared to those with HRV-A and HRV-C infection. Comparisons of controls to those with HRV-A and HRV-C infections showed that, other than for controls with HRV-A being more likely to be LytA positive on whole blood (10% versus 4%, P ϭ 0.012), there were no differences in demographic characteristics, HRV load, and symptoms of RTI (Table 2).
Further, when the HRV types were compared among the asymptomatic and RTI controls separately, there were no significant differences for any of the demographic, clinical, or molecular markers (Table S1 and S2).
Molecular subtyping of the HRV-associated cases. Among the HRV-associated pneumonia cases, 97% of the samples were successfully subtyped, while 3% (n ϭ 11) of the samples failed to amplify, all of which had very low copy numbers (threshold cycle  , with the caveat that maternal exposure must be confirmed by maternal serology for seronegative infants aged Ͻ7 months. c Underweight was defined as weight for age Յ2 SD of the median age-sex-specific WHO reference. d Premature birth was defined as gestational age Ͻ37 weeks. e Tachypnea was defined as respiratory rate Ͼ60 breaths/min for subjects aged Ͻ2 months, respiratory rate Ͼ50 breaths/min for subjects aged 2 to 12 months, and respiratory rate Ͼ40 breaths/min for subjects aged Ͼ12 months. f Fever was defined as temperature Ն38°C. g The child was completely asymptomatic for all signs of respiratory tract illness, including runny nose, fever, cough, wheezing, and difficulty breathing. h CRP was defined as levels Ն40 mg/liter, which are considered to show potential bacterial infection. Only a subset of randomly chosen controls had CRP testing conducted at the South African site. i Blood sample positive for S. pneumoniae colonization by LytA PCR. j HDP was defined as S. pneumoniae density in nasopharynx Ͼ6.9 log 10 copies/ml and/or density in whole-blood sample Ͼ2.2 log 10 copies/ml. k HRV load in the nasopharynx, expressed as log 10 copies/ml. l HRV was the only respiratory virus detected in the nasopharynx. were typed as enterovirus and closely related member of HRV. The species of samples was successfully identified in 91% (n ϭ 415) of the HRV-associated cases, among which 48%, 45%, and 7% were found to be infected with HRV-A, HRV-C, and HRV-B (7%), respectively ( Table 1). The distribution of HRV species differed by age group, with HRV-A (52%) being the most prevalent among infants Ͻ12 months of age, followed by HRV-C (38%) and then HRV-B (10%), whereas HRV-C (60%) was more prevalent among children 13 to 59 months of age, followed by HRV-A (38%) and HRV-B (2%, P ϭ 0.002).
Similarly to the results seen with the controls, HRV-B was the least prevalent species and appeared sporadically throughout the 2-year period (Fig. 1). This limited any in-depth statistical analysis specific to HRV-B, with no significant differences in demographic and clinical characteristics observed between HRV-B-associated cases and HRV-A-associated or HRV-C-associated cases, except that the subjects with HRV-B infection were younger (4.8 months of age) than the subjects with HRV-A infection (9.4 months, P ϭ 0.01) and with HRV-C infection (12.1 months, P Ͻ 0.001) ( Table 3).
Further species-specific analyses were limited to comparing HRV-A-associated cases to HRV-C-associated cases, with HRV-A cases being younger (9.4 months) than HRV-C cases (12.1 months, P ϭ 0.033) cases (Table 3). Furthermore, the HRV-A-associated cases were more likely to have radiographically confirmed pneumonia (abnormal chest radiograph, defined as primary endpoint pneumonia or presence of any infiltrates) than the HRV-C-associated cases (46% versus 36%, P ϭ 0.040) and were more likely to present with concurrent diarrhea (26% versus 14%, P ϭ 0.007). In contrast, cases with HRV-C were more likely to present with wheeze (35%) than cases with HRV-A (25%, P ϭ 0.031) ( Table 3). Among the HRV-associated cases, HRV was the only virus identified in the nasopharynx samples from 54% (n ϭ 223/415) of the cases, including 55%, 42%, and 54% of HRV-A, HRV-B, and HRV-C infections, respectively (P ϭ 0.742).
Case-control comparison of results of molecular subtyping of HRV. There were 60 different HRV-A strains, 17 different HRV-B strains, and 28 different HRV-C strains circulating throughout the 2-year period (Fig. 2). No discernible differences were noted in the distribution of strains between cases and controls; moreover, there were no apparent differences in the distribution of strains among the three sites ( Fig. S2 to S4). Additionally, no obvious patterns of temporal clustering of HRV species or strains were observed, with strain distributions differing on a month-to-month basis (data not shown).
The HRV sequences for each of the species formed many unique clusters, with mean levels of nucleotide diversity of 82% for HRV-A (nucleotide diversity range, 45% to 100%), 80% for HRV-B (nucleotide diversity range, 53% to 100%), and 74% for HRV-C (nucleotide diversity range, 52% to 100%). The levels of nucleotide diversity did not differ among cases and controls for the different HRV species. The HRV-A and HRV-B sequences clustered with the GenBank sequences of known HRV-A and HRV-B strains, with statistically significant bootstrap support, whereas the HRV-C sequences tended to form numerous subclusters which did not always cluster closely with the GenBank sequences but which did always occur in the same monophyletic groups of known HRV-C strains. The numerous clusters and ranges of nucleotide diversities, especially in the HRV-C population, suggest considerable diversity in the strains present in the population (Fig. S2 to S4).
Among the HRV-A strains, the nucleotide similarities to the closest GenBank prototype reference strains ranged from 82.3% to 99.4% and strain identities with other contemporaneous HRV-A strains ranged from 79.8% to 100%. Among the HRV-B strains, the nucleotide similarities to the closest GenBank prototype reference strains ranged from 87.9% to 99.8% and strain identity with other contemporaneous HRV-B strains ranged from 91.2% to 100%. Among the HRV-C strains, the nucleotide similarities to the closest GenBank prototype reference strains showed much lower levels of relatedness than the other two species and ranged from 74.5% to 98.8%; however, there were high degrees of similarity with the other contemporaneous HRV-C strains, with strain identities ranging from 93.9% to 100%. No novel serotypes were identified in this study.

DISCUSSION
This report presents results of genotyping of 836 HRV-positive samples, including 415 from children hospitalized with severe or very-severe pneumonia and 421 from community controls, representing the largest and most in-depth case-control study reporting on HRV molecular subtyping to date. HRV-A was the dominant species identified among cases (48%) and controls (45%), followed closely by HRV-C (45% each among cases and controls), whereas HRV-B was seen only intermittently and accounted for 7% and 10% of HRV strains among cases and controls, respectively.
Among the three sub-Saharan countries, the overall prevalences of HRV detection and viral loads did not differ between cases (21%) and controls (20%). However, presence of HRV in the nasopharynx was associated with case status (21% versus 16% among controls) among children in the age group of 12 to 59 months; among those children, a higher percentage of HRV were HRV-C species in cases (12%) than in controls (7%). Furthermore, HRV loads did not differ significantly between the HRV species, and none of the HRV species were more likely to be associated with either more bacterial coinfections or more viral coinfections regardless of case or control status. The HRV species distributions among cases and controls in our study in both hospitalized and control populations are similar to those reported previously by others (24)(25)(26)(27)(28). Notably, a number of the HRV-positive cases (8%) and controls (7%) were also positive for S. pneumoniae detection in the blood; those levels are comparable to the overall levels of S. pneumoniae detection in the whole of the PERCH study (7% and 6% for cases and controls, respectively) (16). Thus, the pneumococcal PCR had low specificity for diagnosing invasive pneumococcal disease in children, and more work is needed in order to understand the pathophysiology and implications of viral coinfections. , with the caveat that maternal exposure must be confirmed by maternal serology for seronegative infants aged Ͻ7 months. c Underweight was defined as weight for age Յ2 SD of the median age-sex-specific WHO reference. d Premature birth was defined as gestational age Ͻ37 weeks. e Abnormal chest X-ray was defined as radiographically confirmed endpoint pneumonia consolidation or presence of any infiltrates. f Hypoxia was defined as (i) a room air pulse-oximetry reading indicating oxygen saturation at Ͻ90% at the two sites at elevation (Zambia and South Africa) or at Ͻ92% at all other sites or (ii) absence of a room air oxygen saturation reading and child on oxygen. g Tachycardia was defined as heart rate Ͼ160 beats per min (bpm) for subjects aged Ͻ11 months, heart rate Ͼ150 bpm for subjects aged 12 to 35 months, or heart rate Ͼ140 bpm for subjects aged 36 to 59 months. h Tachypnea defined as respiratory rate Ͼ60 breaths/min for subjects aged Ͻ2 months, respiratory rate Ͼ50 breaths/min for subjects aged 2 to 12 months, or respiration rate Ͼ40 breaths/min for subjects aged Ͼ12 months. i Fever was defined as temperature Ͼ38°C. j Leucocytosis was defined as white blood cell count Ͼ15,000 cells/l for subjects aged Ͻ12 months or white blood cell count Ͼ13,000 cells/l for subjects aged Ͼ12 months. k CRP was defined as levels Ն40 mg/ml, which are considered to potentially indicate bacterial infection. l Blood culture positive for any significant noncontaminate bacteria. m Blood sample positive for S. pneumoniae colonization by LytA PCR. n MCPP, S. pneumoniae cultured from a normally sterile body fluid-blood, pleural fluid, or lung aspirate-or pleural fluid or lung aspirate gave a PCR LytA-positive result. o HDP defined as S. pneumoniae density in nasopharynx Ͼ6.9 and/or density in whole-blood sample Ͼ2.2 log10 copies/ml. p HRV load in the nasopharynx (expressed as log 10 copies per milliliter).
Previous smaller studies from Africa, Asia, Europe, America, and Australia (29)(30)(31)(32) have suggested that HRV-C may cause more-severe illness and is more prevalent in cases of lower respiratory tract infections (LRTI) than HRV-A and HRV-B. This was not evident in our study, however, where both HRV-A and HRV-C were ubiquitous throughout the study period in cases and controls and were similarly prevalent among the cases. Additionally, there was no evidence that the cases associated with HRV-C infection had more-severe disease than the cases associated with HRV-A infection and with HRV-B infection, based on presence of hypoxia, presenting as very-severe pneumonia, prolonged hospital stay (Ͼ3 days duration), need for mechanical ventilatory support, or case fatality rate. Instead, among HRV-associated cases, those with HRV-A infection were more likely to have radiographically confirmed pneumonia and concurrent diarrhea than those with HRV-C infection.
That HRV-C infection is more common in older children hospitalized with pneumonia has also been reported by others (25,33,34). It has been suggested previously that the association between HRV-C and older children might be linked to asthma exacerbation and wheezing in these children. Previous studies have reported that HRV-C is more commonly associated with wheezing exacerbation than HRV-A or HRV-B (33)(34)(35)(36). Similarly, in our study, the presence of wheezing was 1.64-fold more common among cases with HRV-C infection than among cases with HRV-A infection. This suggests that some of the children fulfilling our study-specified definition of "pneumonia" might instead have had reactive airway disease following HRV-C infection. This could have been the case despite our study having been designed to exclude likely asthma cases HRV-A strains case control HRV-C strains case control by performance of a bronchodilator nebulizer inhalation challenge prior to study enrollment; i.e., children in whom the lower chest wall indrawing resolved post-B2agonist nebulization challenge, regardless of its effect on wheezing, were not enrolled. These associations of HRV-A with more-severe LRTI and HRV-C with more wheezing disease have also been reported from a study in Burundi, where HRV-A was more prevalent among pneumonia and bronchitis cases and HRV-C among cases with acute wheezing (37). An association between HRV-C and wheezing disease was also seen in a study which characterized the cell receptors for HRV-C infection in humans, namely, the CDHR-3 receptors, which facilitate HRV-C adhesion and replication (16). A mutation (cysteine-to-tyrosine mutation at amino acid 529) in CDHR-3 enhances HRV-C binding and replication in vivo and has also been associated with increased susceptibility to wheezing and asthma illnesses (38).
In our study, HRV-A and HRV-C strains were present throughout the study period, with a highly heterogeneous population of over 100 strains identified. HRV-A, with 60 different strains, had the most diverse genetic population, followed by HRV-C with 28 strains and HRV-B with 17 different strains identified over the study period in both the cases and controls. Furthermore, there were no discernible relationships between the strains identified among cases and controls and no real evidence of temporal clustering of strains over time. However, this study was not sufficiently powered for statistical analysis of case-control status or clinical epidemiology among cases for the different strains making up each of the HRV species. Furthermore, we failed to type 3% (n ϭ 25) of the HRV-positive samples (with similar percentages between cases and controls), which was largely due to PCR failure as a consequence of very low viral loads (C T Ͼ37). The possibility cannot be excluded that these PCR failures might have been due to variability in the primer annealing sites; thus, some genetic variants might have been missed by our typing assays. However, the typing failure rate of 3% is lower than that seen in other studies which looked at HRV serotyping by targeting the VP4-VP2 capsid region (15% to 44% failure) (19,20,26,27). The 5=NCR serotyping technique has been found to have an increased sensitivity for HRV serotype detection in clinical samples compared to the more traditional VP4-VP2 PCR sequencing technique; further, it does not require a nested PCR or multiple primer pairs, thus reducing contamination rates, cost, and time required for serotyping (19).
Regardless, the considerable genetic diversity of HRV reported in this study highlights the heterogeneity of the HRV strains circulating within the general population and further emphasizes the difficulties in attributing causality of disease to HRV. The similarities in strains among cases and controls also negate the assertion that some strains might be more virulent. Other studies have also reported the diverse nature of the HRV genetic population among cases and controls, including studies from South Africa (28), Botswana (27), Kenya (26), and other countries (24,25,39) as well as the lack of obvious seasonal patterns among cases (35,37,(40)(41)(42).
Study limitations included the lack of a gold standard for the determination of the actual cause of the pneumonia episode. Additionally, the cross-sectional design of the study means only a single specimen was taken upon admission to the hospital. The viral load is known to vary with time since onset of illness; thus, a longitudinal study would have allowed us to compare peak viral loads between subjects infected with the different HRV species and between cases with different levels of clinical severity. Furthermore, our study was not designed to analyze the role of HRV in the lower respiratory tract. Sampling the upper respiratory tract is more convenient and less invasive for the patient; however, detection of a pathogen and its viral load in these samples might not reflect the viral burden in the lung parenchyma. Direct sampling of the site of infection, namely, the lung, through lung aspirate sampling or bronchoalveolar lavage would provide more-direct evidence of the role of HRV in pneumonia; however, these sampling techniques are more invasive and difficult to perform in infants and young children. Animal models are needed to further study the pathogenicity of HRV infections.
In conclusion, this report emphasizes that the HRV populations circulating both in children hospitalized with pneumonia and in community controls are highly divergent, that similar strains are circulating over a large geographical location, and that the strains are similar year to year. Additionally, it further highlights the difficulty in attributing causality of LRTI disease to HRV in infants, as no differences were observed in the levels of prevalence of HRV detection between cases and controls (other than in the Ͼ12-month age group) and no differences were found in HRV genotypes between cases and controls, especially among infants. Nevertheless, among the cases, HRV-A tended to be more prevalent among younger children and was associated with more-severe disease and radiographically confirmed pneumonia compared to HRV-C infections, whereas HRV-C was more prevalent among older children with wheezing disease.