Discussion

Our study showed that RSV is an important cause of mild and severe respiratory infections in children under 5 years of age in CAF2,22,25 and is the first to explore in details the epidemiology of RSV in the country. Young infants are usually more often represented in surveillance studies on respiratory infections, due to the immaturity of their immune system. Compared to the other age groups, children aged 0-6 months had an increased risk of being hospitalized regardless of their (RSV) infection status, and the detection rate of RSV was high during the first months after birth — 13.4% in the 0–6 month age group — and decreased with age, to 4.0% in children 4-5 years old. Previous reports have shown higher infection rates during the first year of life linked to primary infections7,12,26-28. Within the first year of life, a significant difference in hospitalization risk according to the month of birth was shown 29, linked to infant age and levels of maternal antibodies elicited by more or less recent RSV exposure of the mother 30,31. RSV infection leads to humoral immune response also in young infants but antibody titers are likely not high enough to confer full protection. This allows reinfections that will then boost humoral immune response 32,33 and explaining why RSV is still detected in older age groups, albeit with lower incidence (4.0-7.5% in CAF). Our analyses also showed that male patients developed more frequently more severe symptoms leading to hospitalisation (33.3% vs 29.6% in female) and were more susceptible to RSV infection (9.5% vs 6.4%), as already reported previously31,34. Hormonal influence on the immune system35 and/or airway anatomical differences31 have been proposed to contribute to sex differences in susceptibility to respiratory infections.
Our study showed an overall prevalence of RSV of 8.0% in the period of 2015-2018, with significant annual (6.4%-10.6%) and seasonal (12.7% in rainy season vs 3.0% in dry season) fluctuations. While RSV circulation is high during the winter in temperate climates12, a high RSV incidence usually coincides with the rainy season in African countries with a tropical climate, such as in Kenya 36, Cameroon 37, Senegal25 and Ghana 38. Also in CAF most cases were reported in November in a study from 201020. RSV seasonality likely depends on climatic factors such as relative humidity, temperature and UV radiation that influence the infectivity of viral particles and stability of aerosols39 or host factors such as overall increased susceptibility to infectious diseases in winter due to lower levels of vitamin D 40-42. Deviations from overall seasonality patterns, as observed in 2017 in CAF where RSV incidence was significantly higher with a peak 3-5 months earlier than usual, have been documented previously. In Germany, an earlier start of the RSV season as compared to the average tends to be linked to higher disease incidence 12. In Switzerland and Finland, seasons with lower incidence and later start alternate with a 2 year-cycle with outbreaks that start earlier in the season 43,44, have a higher incidence and increased hospitalisation rates43. Outbreak periodicity is likely influenced by the level of herd immunity developed during previous seasons and/or RSV genetic diversity. In China, RSV-A dominated seasons started earlier and lasted longer than RSV-B dominated seasons 45. Although RSV incidence in 2016 in CAF did not differ from 2015 (p=0.931) or 2018 (p=0.345), RSV-B was the predominant subtype in that year, likely affecting herd immunity against RSV-A and thus allowing earlier and wider RSV-A circulation in 2017. Long-term RSV surveillance in CAF is needed to better understand the periodicity of RSV-A and B dominance that differs between countries 45-47, and to refine data on RSV season onset, peak and duration in the country.
Phylogenetic analyses, based on partial glycoprotein G sequences, revealed an RSV-A genotype replacement in CAF. While NA1 was the dominant genotype in 2015, its detection rates decreased and it was not found any longer in 2018. RSV temporal genotype replacement is well documented 48-50 and is facilitated by viral evolution and immune selection 51. Notably, RSV-A ON1 and RSV-B BA genotypes, with a 72 or 60 nt duplication located in the second hypervariable region of the G protein 4,19,50, have spread worldwide. They have become predominant across different continents 2,4,19,38,50,52, likely due to a fitness advantage conferred by the duplication 53. In addition, temporal strain replacement within a genotype also occurred in CAF. Within ON1, one large cluster of 79 identical strains in 2017-2018 suggested sustained local transmission. Smaller clusters of identical strains identified during two consecutive seasons (2016-2017 or 2017-2018) or detected in 2016 and 2018 also suggested that these strains were maintained in the country over time. Sporadic cases detected during the dry season, outside the main RSV seasons, may maintain RSV transmission in the population between two outbreaks, without the need for virus re-introduction 54. However, CAF strains interspersed with RSV strains from abroad, suggesting that new RSV strains are also introduced in the country. Maintaining surveillance in CAF as well as in neighbouring countries while increasing the sequencing effort, both concerning the number of strains and the sequence length, will help to characterize the importance of local versus imported strain circulation in the country.
While the NA1 genotype detected in CAF showed limited polymorphism14,55 potentially contributing to its elimination in CAF 12, the ON1 and BA9 CAF strains showed a higher degree of polymorphism. Among the 33 substitutions observed in the ON1 CAF strains compared to the prototype strain, L274P substitution has been linked to the RSV evasion from antibodies 2. Among the 23 mutations in BA9 strains, the genotype-specific substitutions L223P, S247P, I281T and H287Y 56,57 were also identified in CAF strains. Gain (D245N substitution in 3 unique NA1 and D273N in 3 unique BA9 strains) or loss (positions 318-320 in 4 unique ON1 strains, 296-298 or 310-312 in 3 BA9 strains) of N-linked glycosylation might potentially affect antigenicity58,59.
Clinical manifestations of RSV infections vary and can include symptoms of both upper and lower respiratory infections 60. In our study, RSV was significantly associated with dyspnea, wheezing, chest indrawing and inability to feed as reported before26, while its association with fever or cough could not be assessed due to the use of WHO case definitions developed for influenza surveillance, constituting a limitation of our study25. Indeed, a substantial number of RSV infected patients, especially young infants 26,61, do not develop fever, while this symptom is part of the ILI and SARI case definitions. Using ARI and extended SARI definitions should increase sensitivity for RSV case identification 26,61. Moreover, screening efficiency can be greatly improved by using real-time RT-PCR. While similar detection rates were found in the USA (6.1%, 62) or in Kenya (8%; 36) when using conventional RT-PCR, much higher detection rates were reported in Germany (23%; 12) and in Ghana (23%;38), when using more sensitive real-time RT-PCR assays63. Therefore, combining revised RSV surveillance criteria and a more sensitive real-time RT-PCR screening approach will improve the sensitivity of RSV detection in the country in the future.