A unified global genotyping framework of dengue virus serotype-1 for a stratified coordinated surveillance strategy of dengue epidemics

Abstract

Background: Dengue is the fastest spreading arboviral disease, posing great challenges on global public health. A reproduceable and comparable global genotyping framework for contextualizing spatiotemporal epidemiological data of dengue virus (DENV) is essential for research studies and collaborative surveillance.

Methods: Targeting DENV-1 spreading prominently in recent decades, by reconciling all qualified complete E gene sequences of 5003 DENV-1 strains with epidemiological information from 78 epidemic countries/areas ranging from 1944 to 2018, we established and characterized a unified global high-resolution genotyping framework using phylogenetics, population genetics, phylogeography, and phylodynamics.

Results: The defined framework was discriminated with three hierarchical layers of genotype, subgenotype and clade with respective mean pairwise distances 2-6%, 0.8-2%, and ≤ 0.8%. The global epidemic patterns of DENV-1 showed strong geographic constraints representing stratified spatial-genetic epidemic pairs of Continent-Genotype, Region-Subgenotype and Nation-Clade, thereby identifying 12 epidemic regions which prospectively facilitates the region-based coordination. The increasing cross-transmission trends were also demonstrated. The traditional endemic countries such as Thailand, Vietnam and Indonesia displayed as persisting dominant source centers, while the emerging epidemic countries such as China, Australia, and the USA, where dengue outbreaks were frequently triggered by importation, showed a growing trend of DENV-1 diffusion. The probably hidden epidemics were found especially in Africa and India. Then, our framework can be utilized in an accurate stratified coordinated surveillance based on the defined viral population compositions. Thereby it is prospectively valuable for further hampering the ongoing transition process of epidemic to endemic, addressing the issue of inadequate monitoring, and warning us to be concerned about the cross-national, cross-regional, and cross-continental diffusions of dengue, which can potentially trigger large epidemics.

Conclusions: The framework and its utilization in quantitatively assessing DENV-1 epidemics has laid a foundation and re-unveiled the urgency for establishing a stratified coordinated surveillance platform for blocking global spreading of dengue. This framework is also expected to bridge classical DENV-1 genotyping with genomic epidemiology and risk modeling. We will promote it to the public and update it periodically.

Keywords: Dengue virus serotype-1 (DENV-1); Global genotyping framework; Molecular epidemiology; Molecular surveillance; Phylogeography; Population structure.

Fig. 1

Global genotyping framework of DENV-1 established based on its population structure, phylogeny, and epidemiology. a, e The dots at the tip of each strains are marked with different colour representing 41 subgenotypes identified. a The unified global framework of DENV-1 subgenotypes is labeled in the phylogenetic tree of the complete E gene sequences of 5003 DENV-1 strains worldwide. Due to inadequate strains recorded, only 5 and 3 strains of genotype II and III were defined in a single subgenotype (2A and 3A) and clade (2A1 and 3A1), respectively. b The defined genotyping framework of DENV-1 represents an increasing intra-taxon divergence of the mean pairwise distances (MPDs). c The correlation between the substitution rate and time to the most common ancestor (tMRCA) was assessed using a linear regression model, after adjusting the inclusive number of strains, MPDs, and Simpson Index. d The dynamics description of the estimated viral population size of DENV-1 with the cumulative percentages of inclusive strains, clades, subgenotypes of DENV-1, and the involving epidemic countries or areas. e The temporal phylogeny of DENV-1 using 910 strains selected among the total 5003 ones worldwide which cover the full spatiotemporal ranges and represent all subgenotypes and clades in genotype I, IV, and V of DENV-1. The features of the selected 910 E complete sequences are presented in Additional file 3: Appendix-Data S-C1. The linear regression of root-to-tip genetic divergences versus strain sampling dates is displayed in Additional file 1: Fig. S C1. The gray vertical dotted lines show three sharp increases in the estimated viral population size of DENV-1 during the three periods of 1976–1977, 1996–1998, and 2005–2006

Fig. 2

Geographical clustering and classification of the twelve epidemic regions of DENV-1 worldwide by countries/areas at the subgenotype level. a Global distribution of the twelve epidemic regions of DENV-1. The pies are sized to indicate the number of strains and the slices are colored by subgenotypes. WAFR, West African Region; RSR, Red Sea Region; SWIO, Southwest Indian Ocean; SASC, South Asia Subcontinent; GMS-China, Great Mekong Subregion-China; SEA, Southeast Asia; PHI, Philippines; TWP, Tropical Western Pacific; OCE, Oceania; CNA, Central North America; CAR, Caribbean; SA, South America. b Left panel: heatmap of distribution of subgenotypes of DENV-1 represented by countries/areas in the defined 12 epidemic regions, and the 32 countries/areas recorded with more than 10 strains subject to the preliminary hierarchical geographical clustering using partitioning around medoid (PAM) in Additional file 1: Fig. S-E3 highlighted in bold fonts. Right panel: the proportion of involving epidemic regions counted by the number of strains in each subgenotypes of DENV-1

Fig. 3

Global transmission patterns of DENV-1 at the subgenotype level. a The transmission patterns assessed by phylogeographical inferences using the Bayesian stochastic search variable selection (BSSVS). 29 subgenotypes of DENV-1 and 67 involving epidemic countries/areas were included, under the condition of the subgenotypes of DENV-1 with > 20 strains as well as the viral lineage transmission with the binary indicator (I) > 0.50 and Bayes factor (BF) > 6. The pies represent the country/area circled with the coloured composition of subgenotypes involving in transmission. The gray dotted oval lines show the twelve belonging epidemic regions. The names of countries/areas with ≥ 5 transmission pairs are shown in red, and ≥ 10 pairs in bold red. WAFR, West African Region; RSR Red Sea Region; SWIO, Southwest Indian Ocean; SASC, South Asia Subcontinent; GMS-China, Great Mekong Subregion-China; SEA Southeast Asia; PHI, Philippines; TWP, Tropical Western Pacific; OCE, Oceania; CNA, Central North America; CAR, Caribbean; SA, South America. b The alluvial plot indicates the subgenotype flow of DENV-1 by travelers. 329 isolates of DENV-1 were recorded from returning travelers. The importing countries and visited countries were confirmed through checking information in Genbank and corresponding references. There were 30 visited countries/areas and 7 importing countries/areas in total. The inner pairs of the alluvial plots show the subgenotype flow of 329 isolates from the visited (exported) countries/areas (Left) to the imported (Right), and the width of alluvial segments indicates the number of isolates. The outer pairs show the subgenotypes proportions of DENV-1 excluded data from travelers in the involving exported (Left) and importing (Right) countries/areas

Fig. 4

Global diffusion patterns of DENV-1 at the clade level inferred by phylodynamics analysis using posterior analysis of coalescent trees (PACT). a The involving epidemic country/area of DENV-1 with a trunk probability > 50% in certain clade in at least 12 consecutive months inferred by PACT, was considered as the dominant epidemic location of this clade. So 31 PACT-inferred dominant countries/areas epidemic with 47 clades of DENV-1 were estimated, represented with colour horizontal lines in the panel. Among them, 1D2, 4E8, 5N7, and 5P8 respectively broke out only in the single country-Vietnam, Indonesia, Brazil, and Colombia. The gray dots show the earliest and latest years of observed isolation for each clade of DENV-1, connected by thin gray lines. b The vertical bars show the number of migrant events in the 25 PACT-inferred dominant epidemic countries or areas when the migration rate was estimated > 0.1. Herein, six countries including Japan, Haiti, Paraguay, Saint Barthelemy, Colombia, and Ecuador were excluded due to the migration rate < 0.1. c The alluvial plot shows the diffusion of 42 clades (excluding 1D2, 4E8, 5N7, and 5P8 in a single country, and 1I1 in Vietnam and Cambodia) from the 25 PACT-inferred dominant countries/areas to 43 countries/areas in 9 epidemic regions, with the width of line indicating the value of migration rate

Fig. 5

Dynamics and trends of the epidemic trunks of the 47 clades of DENV-1 from 2005 to 2015 and the involved PACT-inferred 31 dominant epidemic countries/areas. The data inferred by PACT showed in Fig. 4 were further subjected to time series dynamics analyses and visualized with 0.1 year step ward. a The proportion of epidemic trunks of the 47 clades of DENV-1 is changed dynamically through time. b The proportion of trunks of the involved PACT-inferred dominant epidemic countries/areas is changed dynamically through time