Genomic epidemiology of the SARS-CoV-2 epidemic in Brazil – Nature.com

COVID-19 transmission dynamics in Brazil

The first confirmed infection of SARS-CoV-2 in Brazil was on 26 February 2020 in the State of So Paulo (SP), in a traveller returning from Italy (Fig. 1a). On 17 March 2020, the first COVID-19-related death, a 61-year-old male, was reported in the same state4,5. Four days later, all Brazilian states reported at least one confirmed case of COVID-19 and the Brazilian Ministry of Health (BRMoH) declared an outbreak of large-scale community transmission of the virus6. By 10 April 2020, the virus had already reached remote locations, such as the Yanomami indigenous community located in the state of Roraima in northern Brazil6 (Fig. 1a).

a, Timeline of SARS-CoV-2 key events in Brazil. b, Epidemic curve showing the progression of reported daily viral infection numbers in Brazil from the beginning of the epidemic (grey) and deaths (red) in the same period, with restriction phases indicated by the horizontal bar at the bottom. c, Map of cumulative SARS-CoV-2 cases per 100,000 inhabitants in Brazil up to June 2021.

After the World Health Organization (WHO) declared the outbreak of SARS-CoV-2 as a public health emergency of international concern on 30 January 2020, the Brazilian government introduced restriction measures to mitigate viral spread (Fig. 1a)7. The primary measure involved social isolation, followed by the closure of schools, universities and non-essential businesses8. Additional measures included the mandatory use of personal protective masks9, the cancellation of events expected to attract large numbers of people and tourists, and opening only of services considered as essential such as markets and pharmacies8,10. However, while the epidemic was growing, restriction measures were progressively eased to mitigate negative impacts on the economy. Notably, even during periods of restriction, travel between Brazilian states largely remained possible, enabling SARS-CoV-2 transmission throughout the country11. Travel was probably linked to the emergence of more contagious viral lineages, such as VOC Gamma (lineage P.1) and VUM Zeta (lineage P.2). Notably, these variants may have contributed to a second wave that was more severe in terms of infections and deaths than the first wave (Fig. 1b)11,12,13,14.

The COVID-19 death toll in Brazil rose steadily after March 2021. It reached a daily total of 4,250 deaths on April 2021, the highest number of daily fatalities from COVID-19 worldwide (Fig. 1b). Signs of collapse of the health system were reported in numerous cities around the country. The situation worsened after multiple VOCs and VUMs emerged during a slow vaccination campaign15. Vaccination in Brazil began on 17 January 2021, when the Instituto Butantan imported the first 6 million doses of CoronaVac (a whole-virus inactivated vaccine) from Sinovac Biotech (Fig. 1a)16,17. As of 16 February 2022, approximately 71.8% of the Brazilian population had been vaccinated with the first dose of any of the vaccines available (CoronaVac, AstraZeneca, Pfizer and Janssen), but only 22% were fully vaccinated (with a single dose of Janssen or two doses of any other vaccine)18.

By analysing the total number of COVID-19 notified cases to the end of September 2021, we observed that the Brazilian region with the highest population density (Southeast) also contained the highest number of the cases registered in the country, with the state of So Paulo documenting the largest number of cases (n=4,369,410) in that period (Fig. 1c). However, when we considered the incidence rate (number of reported cases per population) by state, we found that the Midwest, the least populated region in Brazil, had the highest incidence rate, with 13,604.23 cases per 100,000 inhabitants1.

A total of 3,866 near-full genome sequences from SARS-CoV-2 RTqPCR positive samples were obtained as part of this study. SARS-CoV-2 sequencing spanned February 2020 to June 2021, with samples from 8 of the 27 Brazilian states (So Paulo, 3,309; Rio Grande do Sul, 48; Paran, 55; Minas Gerais, 80; Mato Grosso do Sul, 36; Mato Grosso, 51; Bahia, 224) and one neighbouring country, Paraguay (n=63). Almost half of the sequences were from Southeast Brazil, comprising the states of So Paulo and Rio de Janeiro that reported the most cases (Fig. 1c)6. Sequenced genomes were from samples collected from 2,023 females and 1,843 males (Supplementary Tables 1 and 2), with a median age of 41.72 years (range: 1 to 90 years of age). All tested samples contained sufficient viral genetic material (2ngl1) for library preparation. For positive samples, PCR cycle threshold (Ct) values were on average 19.93 (range: 10.7530). Sequences had a median genome coverage of 95% (range: 8099.99) and average genome coverage was typically higher for samples with lower Ct values (Supplementary Fig. 1). Epidemiological information and sequencing statistics of the generated sequences from Brazil and Paraguay are reported in Supplementary Tables 1 and 2, respectively. Sequences were assigned to 39 different PANGO lineages on the basis of the proposed dynamic nomenclature for SARS-CoV-2 lineages (Supplementary Fig. 1, and Tables 1 and 2) and have been submitted to GISAID following the WHO guidelines (Supplementary Tables 1 and 2) (Pangolin version 3.1.7, August 2021).

The rapid spread of SARS-CoV-2, together with the reported circulation of several VOCs and VUMs in Brazil, prompted an intensification of genomic surveillance by the National Network for Pandemic Alert of SARS-CoV-2 at the end of December 2020. As of 30 June 2021, more than 17,135 SARS-CoV-2 genomes from all 27 Brazilian states had been deposited in the GISAID database (Fig. 2a). The states with the highest number of sequenced genomes were So Paulo (n=9,600) and Rio de Janeiro (n=2,031). Although genomic surveillance began as soon as the first confirmed infections were detected in Brazil, by the end of June 2021 there was still a paucity of genomic data from some states, such as Roraima (n=29), Acre (n=29), Rondnia (n=37), Tocantins (n=27), Piau (n=19) and the Federal District (n=33) (Fig. 2a). Half of all Brazilian genomes were deposited in early 2021, suggesting that surveillance was at its peak in the second wave following the emergence of Gamma (and other VOCs (for example, Alpha/B.1.1.7)) and VUMs (for example, Zeta) throughout the country (Fig. 2b).

a, Map of Brazil with the number of sequences in GISAID as of 30 June 2021. The map is coloured according to geographical macro region: North (red), Northeast (green), Southeast (purple), Midwest (light blue) and South (light orange). AC, Acre; AL, Alagoas; AP, Amap; AM, Amazonas; BA, Bahia; CE, Cear; DF, Distrito Federal; ES, Esprito Santo; GO, Gois; MA, Maranho; MT, Mato Grosso; MS, Mato Grosso do Sul; MG, Minas Gerais; PA, Par; PB, Paraba; PR, Paran; PE, Pernambuco; PI, Piau; RR, Roraima; RO, Rondnia; RJ, Rio de Janeiro; RN, Rio Grande do Norte; RS, Rio Grande do Sul; SC, Santa Catarina; SP, So Paulo; SE Sergipe; TO, Tocantins. b, Temporal sampling of sequences in Brazilian states through time with VOCs highlighted and annotated according to their PANGO lineage assignment. c, Time-resolved maximum-likelihood phylogeny containing high-quality near-full genome sequences from Brazil (n=3,866) obtained from this study, analysed against a backdrop of global reference sequences (n=25,288). VUMs and VOCs are highlighted on the phylogeny. d, Sources of viral introductions into Brazil characterized as external introductions from the rest of the world. e, Sources of viral exchanges (imports and exports) into and out of Brazil. f, Number of viral exchanges within Brazilian regions by counting the state changes from the root to the tips of the phylogeny in c.

To understand the dynamics of SARS-CoV-2 spread in Brazil, we coupled epidemiological data with phylodynamic analysis for a data set comprising 25,288 available globally representative genomes, including the genomes sequenced in this study (n=3,866) sampled from 26 December 2019 to 28 June 2021 (Figs. 2c and 3). A date-stamped phylogeny of these data indicated that most of the Brazilian sequences were interspersed with those introduced from several countries (Figs. 2c,d). This pattern further indicated that the co-circulation of multiple SARS-CoV-2 lineages over time was linked to multiple importations followed by large local transmissions concomitant with a high number of infections (Fig. 2c,d).

Time-resolved maximum-likelihood phylogeny containing 17,135 high-quality Brazilian SARS-CoV-2 near-full-genome sequences (n=3,866 generated in this study) analysed against a backdrop of global reference sequences. VUMs and VOCs are highlighted.

Using an ancestral location state reconstruction on the dated phylogeny, we were able to infer the number of viral imports and exports between Brazil and the rest of the world, and between individual Brazilian regions (hereafter referred as the North, Northeast, Midwest, Southeast and South regions) (Fig. 2df). The bulk of imported introductions (estimated to be 114 independent ones) were largely from Europe (Fig. 2d), occurring before the implementation of restriction measures (April 2020) when the epidemic was rapidly progressing (Fig. 2d,e). However, at least 33 introduction events were inferred to have occurred during enforcement of preventive measures up to August 2020 (Fig. 2d,e), and hence before those measures were loosened. Finally, although Brazil was a major virus importer, there were approximately 10 times more inferred exportation events out of Brazil than viral introductions into Brazil (Fig. 2e).

Our estimates of viral movement within Brazil further suggested that the Southeast region was the largest contributor of viral exchanges to other regions, comprising approximately 40% of viral movements from one geographical region to another, followed by the North region that contributed to approximately 25% of all viral movements. Although these estimates are in line with epidemiological data, this observation is probably also influenced by these two regions having the greatest number of sequences available for analysis.

We next focused on two variants (Gamma/P.1 and Zeta/P.2) that evolved from the B.1.1.28 lineage and grew into large transmission clusters during the second wave of the epidemic in Brazil after January 2021. To assess the detailed evolution of these lineages over time, we performed a spatiotemporal phylogeographic analysis using a molecular clock model.

The Gamma VOC was first sampled in Brazil in early January 202112,19. It displayed an unusual number of lineage-defining mutations in the S protein, including three designated that may impact transmission, immune escape and virulenceN501Y, E484K and K417T20,21,22. In line with previous estimates12,19, our phylogeographic analysis suggested that the Gamma variant emerged around 21 November 2020 (95% highest posterior density, 1229 November 2020) in Manaus (Amazonas state) in Northern Brazil and spread extensively among Brazilian regions (Fig. 4a,c). Our data reveal multiple introductions of this lineage from the Amazonas state to Brazils southeastern, northeastern and midwestern states (Fig. 4a,c). By mid-January 2021, the southeastern and northern regions had also acted as source populations for the introduction of this variant into the neighbouring southern region (Fig. 4a,c).

a, Phylogeographic reconstruction of the spread of the Gamma VOC in Brazil. Circles represent nodes of the maximum clade credibility phylogeny and are coloured according to their inferred time of occurrence. Shaded areas represent the 80% highest posterior density interval and depict the uncertainty of the phylogeographic estimates for each node. Solid curved lines denote the links between nodes and the directionality of movement. Differences in population density are shown on a dark-white scale. b, Phylogeographic reconstruction of the spread of the Zeta VUM across Brazil. Circles represent nodes of the maximum clade credibility phylogeny and are coloured according to their inferred time of occurrence. Shaded areas represent the 80% highest posterior density interval and depict the uncertainty of the phylogeographic estimates for each node. Solid curved lines denote the links between nodes and the directionality of movement. Differences in population density are shown on a dark-white scale. c, Number of exchanges of the Gamma variant between Brazilian regions (N, North; NE, Northeast; MD, Midwest; SE, Southeast; S, South). d, Number of exchanges of the Zeta variant between Brazilian regions. e, Sources of viral export of the VOC and VUM from Brazil to the rest of the world.

Zeta (P.2) is defined by the presence of the S:E484K mutation in the receptor binding domain (RBD) and other lineage-defining mutations outside the S protein13,14. Although it was first described in samples from October 2020 in the state of Rio de Janeiro, our phylogeographic reconstruction suggests that the variant originated from Paran state in South Brazil in late August 2020 (95% highest posterior density, 19 August to 03 September 2020) (Fig. 4b). Since then, Zeta has spread multiple times to much of the southeastern, northeastern, midwestern and northern Brazilian regions (Fig. 4d). Together, our results further suggest that the transmission dynamics roughly followed patterns of population density, moving most often between the most populous localities (Fig. 4a,b).

By estimating the pattern of migration flows, we also examined the potential role of Brazil as an exporter of the Gamma and Zeta variants to the rest of the world (Fig. 4). While the North region seeded approximately 47% of all Gamma infections into other regions, consistent with it being where this lineage originated, there is strong evidence both from phylogeographic analysis (Fig. 4a) and ancestral state reconstruction (Fig. 4c) that there was considerable subsequent transfer of Gamma between all regions. Zeta had a different dispersal pattern from Gamma, with 73% of all Zeta movements originating from the Southeast and South regions, consistent with our phylogeographic reconstruction that this is the geographic source of this lineage (Fig. 4).

Our analysis further revealed that Brazil has contributed to the international spread of both variants, with at least 316 and 32 exportation events to the rest of the world detected for Gamma and Zeta variants, respectively (Fig. 4e). Consistent with importations, most exports were to South America (65%) and Europe (14%), followed by Asia (11%), North America (5%), Africa (2.5%) and Oceania (2.5%), with an increase between January and March 2021 coinciding with the second wave of infections in Brazil and some relaxation of international travel restrictions (Fig. 4e). As shown elsewhere, these results demonstrate that under relaxation of travel restrictions, SARS-CoV-2 lineages can spread to a diverse range of international locations23,24,25,26,27,28.

To explore the burden of the Brazilian SARS-CoV-2 pandemic on other South American countries, we provide a preliminary overview of the SARS-CoV-2 epidemic in Paraguay. The first COVID-19 confirmed case was documented in Paraguay on 7 March 2020 in a 32-year-old man from San Lorenzo, Central Department. Thirteen days later, the first death and the first case of community transmission were also confirmed. COVID-19 cases in Paraguay rose sharply in March (Fig. 5a), resulting in 100% occupancy of intensive care beds, prompting the government to declare a strict quarantine to mitigate the spread of the virus29,30. By the end of June 2021, a total number of 460,000 confirmed cases and 15,000 coronavirus-related deaths had been reported in Paraguay29.

a, Epidemic curve showing the progression of reported viral infection numbers in Paraguay from the beginning of the epidemic (grey) and deaths (red) in the same period. b, Progressive distribution of the top 20 PANGO lineages in Paraguay over time. c, Time-resolved maximum-likelihood tree containing high-quality near-complete genome sequences from Paraguay (n=63) obtained in this study, analysed against a backdrop of global reference sequences. VUMs and VOCs are highlighted on the phylogeny. Genome sequences from Paraguay obtained in this study are highlighted with red borders.

The COVID-19 epidemic in Paraguay can generally be characterized by three phases: phase I starting from 10 March 2020, characterized by restriction measures; phase II since 4 May 2020, also called intelligent/smart quarantine with a gradual return to work and social activities; and phase III implemented since 5 October 2020, known as the COVID way of living, characterized by the relaxation of the restriction measures and the reopening of national borders and resumption of international flights30.

Since the beginning of the epidemic, there has been a paucity of whole-genome sequences from Paraguay, with only n=165 whole-genome sequences available on GISAID by the end of July 2021, about 0.0003% of known cases. This seriously impacts the ability to characterize the molecular epidemiology of SARS-CoV-2 at a regional level. In collaboration with the Pan-American Health Organization and the National Public Health Laboratory of Asuncin in Paraguay, we obtained a total of 63 near-complete genome sequences sampled between July 2020 and June 2021, representing ~40% of the currently available genomes from this country. The selection of the samples was based on the Ct value (30) and availability of epidemiological metadata, such as date of sample collection, sex, age and municipality of residence. Thus, by applying these inclusion criteria, only 63 positive samples were considered suitable for this study. As expected, we observed the co-circulation of multiple SARS-CoV-2 lineages (Fig. 5b), linked to multiple importations and subsequently characterized by large transmission clusters.

Importantly, our phylogenetic analysis revealed that most of the SARS-CoV-2 variants currently circulating in Paraguay, including lineages B.1.1.28, B.1.1.33, Zeta and Gamma, originally emerged in Brazil (Fig. 5a,b), thus suggesting cross-border transmission from Brazil to Paraguay (Fig. 5b,c). This reinforces the importance of non-pharmaceutical measures in containing and preventing the spread of viral strains into neighbouring countries.

As of 31 July 2021, a total of 78% of available genomic sequences from Paraguay were linked to infections caused by Brazilian variants, with the Gamma VOC being the most prevalent lineage in the country. As genome sequencing is not widespread, it is difficult to determine how widely these variants have spread within Paraguay and to other Latin American countries. However, the abundance of COVID-19 cases in Brazil, a country that shares borders with ten countries, suggests that this risk is probably high.

Read more here:
Genomic epidemiology of the SARS-CoV-2 epidemic in Brazil - Nature.com