Successive Respiratory Syncytial Virus Epidemics in Local Populations Arise from Multiple Variant Introductions, Providing Insights into Virus Persistence

Abstract

Respiratory syncytial virus (RSV) is a global respiratory pathogen of humans, with infection occurring characteristically as recurrent seasonal epidemics. Unlike influenza viruses, little attention has been paid to the mechanism underlying worldwide spread and persistence of RSV and how this may be discerned through an improved understanding of the introduction and persistence of RSV in local communities. We analyzed 651 attachment (G) glycoprotein nucleotide sequences of RSV B collected over 11 epidemics (2002 to 2012) in Kilifi, Kenya, and contemporaneous data collected elsewhere in Kenya and 18 other countries worldwide (2002 to 2012). Based on phylogeny, genetic distance and clustering patterns, we set out pragmatic criteria to classify local viruses into distinct genotypes and variants, identifying those newly introduced and those locally persisting. Three genotypes were identified in the Kilifi data set: BA (n = 500), SAB1 (n = 148), and SAB4 (n = 3). Recurrent RSV epidemics in the local population were composed of numerous genetic variants, most of which have been newly introduced rather than persisting in the location from season to season. Global comparison revealed that (i) most Kilifi variants do not cluster closely with strains from outside Kenya, (ii) some Kilifi variants were closely related to those observed outside Kenya (mostly Western Europe), and (iii) many variants were circulating elsewhere but were never detected in Kilifi. These results are consistent with the hypothesis that year-to-year presence of RSV at the local level (i.e., Kilifi) is achieved primarily, but not exclusively, through introductions from a pool of variants that are geographically restricted (i.e., to Kenya or to the region) rather than global.

Importance: The mechanism by which RSV persists and reinvades local populations is poorly understood. We investigated this by studying the temporal patterns of RSV variants in a rural setting in tropical Africa and comparing these variants with contemporaneous variants circulating in other countries. We found that periodic seasonal RSV transmission at the local level appears to require regular new introductions of variants. However, importantly, the evidence suggests that the source of new variants is mostly geographically restricted, and we hypothesize that year-to-year RSV persistence is at the country level rather than the global level. This has implications for control.

FIG 1

Temporal patterns and genetic diversity in RSV strains from Kilifi 2002-12. (A) Combined monthly detection frequency of RSV A and B viruses from both the KCH child inpatient surveillance study, 2002-12, and the community birth cohort study, 2002-05. The bars show RSV B detection (black, hospital study; brown, cohort study). The gray show RSV A detection (dotted line, hospital study; solid line, cohort study). (B) ML phylogenetic tree showing the relationship of all 651 sequences. The Kilifi sequences formed three major clusters: SAB1 (green), BA (red), and SAB4 (blue). (C) Monthly occurrence of the 3 genotypes from January 2002 to June 2012.

FIG 2

Clustering of the group B viruses identified in Kilifi (2002-12) to define genotype and variant. (C) The ectodomain region of Kilifi viruses was analyzed with the USEARCH algorithm. Thresholds corresponding to genotype (0.965) and variant (0.99), i.e., corresponding to the 4-nucleotide definition (see Materials and Methods) are indicated.

FIG 3

ML phylogenetic trees of G sequences from the individual Kilifi epidemics. (A) Genotype SAB1 viruses detected in 4 epidemics (2001-02 viruses were combined with 2002-03 sequences in panel i). (B) Genotype BA viruses detected in 10 epidemics (epidemic 2002-03 and epidemic 2003-04 viruses were combined in panel i). The trees include only unique sequences for each epidemic. Variant names are next to the phylogenetic clusters, e.g., SAB1-V1 for SAB1 variant 1. A repeated variant name implies persistence between epidemics. Numbers in parentheses are numbers of other sequences that were identical to the sequence shown.

FIG 4

BEAST tree showing the phylogenetic and temporal placement of the 282 unique Kilifi sequences reported in this study. The taxon names are colored by epidemic: black, 2001-02; cyan, 2002-03; pink, 2003-04; maroon, 2004-05; light blue, 2005-06; green, 2006-07; orange, 2007-08; dark green, 2008-09; bright green, 2009-10; blue, 2010-11; red, 2011-12. Variant names that were derived in Fig. 3 are indicated next to the sequence clusters or singletons. The node bars indicate the 95% highest posterior density (HPD) height interval, while the sizes of the filled circles on the nodes indicate the level of posterior support for the associated branch. Variant names in gray indicate those observed on multiple branches during the same epidemic.

FIG 5

Radial ML phylogenetic trees showing the incremental genetic relatedness of each new epidemic strains sequences relative to the previous epidemic sequences from Kilifi. (A) Genotype SAB1 viruses detected in 4 epidemics (2001-02) viruses (open red circles) and 2002-03 epidemic viruses are combined in panel i. (B) Genotype BA viruses detected in 10 epidemics (2002-03, 2002-04, and 2004-05 viruses are combined in panel i). Throughout the trees, the strains that were perceived to be of newly introduced variants are represented by a red symbol; those perceived to belong to variants that persisted from previous Kilifi epidemics are represented by a green symbol. Tips without a marker imply that the associated strains had been observed in previous epidemics but were not observed in the current epidemic.

FIG 6

Global context of the Kilifi group B viruses. (A) ML phylogenetic tree comparing Kilifi 2002-12 unique sequences (n = 282) and contemporaneous data from around Kenya and 18 other countries (n = 629). The taxon names of Kenyan sequences are in red for Kilifi, blue for Nairobi, green for Dadaab, and pink for Nyanza. Taxon names of sequences from outside Kenya (background diversity) are in black. (B) Example of a cluster without a Kenyan sequence. (C) Example of a cluster with a majority of Kenyan sequences but with a few sequences from abroad, as well as supporting global migration of the virus. (D) An example of a cluster of Kilifi sequences intermixing with sequences from other parts of Kenya to show case-within-country virus migration and perhaps epidemic seeding.