Editor's evaluation: Genomic epidemiology of the first two waves of SARS-CoV-2 in Canada
Year: 2021
Type: peer-review
Abstract: Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Tracking the emergence and spread of SARS-CoV-2 lineages using phylogenetics has proven critical to inform the timing and stringency of COVID-19 public health interventions. We investigated the effectiveness of international travel restrictions at reducing SARS-CoV-2 importations and transmission in Canada in the first two waves of 2020 and early 2021. Maximum likelihood phylogenetic trees were used to infer viruses’ geographic origins, enabling identification of 2263 (95% confidence interval: 2159–2366) introductions, including 680 (658–703) Canadian sublineages, which are international introductions resulting in sampled Canadian descendants, and 1582 (1501–1663) singletons, introductions with no sampled descendants. Of the sublineages seeded during the first wave, 49% (46–52%) originated from the USA and were primarily introduced into Quebec (39%) and Ontario (36%), while in the second wave, the USA was still the predominant source (43%), alongside a larger contribution from India (16%) and the UK (7%). Following implementation of restrictions on the entry of foreign nationals on 21 March 2020, importations declined from 58.5 (50.4–66.5) sublineages per week to 10.3-fold (8.3–15.0) lower within 4 weeks. Despite the drastic reduction in viral importations following travel restrictions, newly seeded sublineages in summer and fall 2020 contributed to the persistence of COVID-19 cases in the second wave, highlighting the importance of sustained interventions to reduce transmission. Importations rebounded further in November, bringing newly emergent variants of concern (VOCs). By the end of February 2021, there had been an estimated 30 (19–41) B.1.1.7 sublineages imported into Canada, which increasingly displaced previously circulating sublineages by the end of the second wave.Although viral importations are nearly inevitable when global prevalence is high, with fewer importations there are fewer opportunities for novel variants to spark outbreaks or outcompete previously circulating lineages. Editor's evaluation This study provides important observations about the transmission of SARS-CoV-2 lineages within Canada and the importation of lineages into Canada over the first year of the COVID-19 pandemic. This information is critical for understanding SARS-CoV-2 evolution and epidemiology, including the potential impacts of travel restrictions. https://doi.org/10.7554/eLife.73896.sa0 Decision letter Reviews on Sciety eLife's review process Introduction The COVID-19 pandemic has highlighted the importance of genomic epidemiology in deciphering the origin and spread of SARS-CoV-2 lineages across local and global scales to aid in directing responses (Deng et al., 2020; Geoghegan et al., 2020; Gonzalez-Reiche et al., 2020; du Plessis et al., 2021; Worobey et al., 2020). Studying evolutionary relationships between SARS-CoV-2 sequences over time and space allows estimation of the relative contributions of international and domestic transmission in association with public health interventions. Thus, phylogenetics can be highly informative in evaluating the effectiveness of non-pharmaceutical interventions at curbing importations and reducing transmission. Phylogenetic analyses are dependent upon timely generation and sharing of publicly available genetic sequences and associated metadata, such as through the Global Initiative on Sharing All Influenza Data (GISAID) platform (Khare et al., 2021; Shu and McCauley, 2017). The global availability of SARS-CoV-2 genomes has been unprecedented, such that by 15 June 2021, just over a year and a half since the first whole genome sequence was shared (de Maio, 2020; Wu et al., 2020), there were 495,159 SARS-CoV-2 sequences on GISAID representing 108,301,802 COVID-19 diagnoses (Krispin and Byrnes, 2020). Nomenclature systems to partition viral sequences sharing common mutations and recent ancestry, such as Pango lineages, have provided a dynamic system for genomic SARS-CoV-2 surveillance (O’Toole et al., 2021; Rambaut et al., 2020). Tracking the emergence, dispersal, and genomic characteristics of lineages, particularly variants of concern (VOCs) and interest (VOIs), has become critical in light of their demonstrated increased transmissibility, potential increased clinical severity, and ability to circumvent host immune responses (Lopez Bernal et al., 2021; Faria et al., 2021; Meng et al., 2021; Planas et al., 2021; Tegally et al., 2020). Therefore, identifying and characterizing predominantly circulating lineages and sublineages is a cornerstone of global epidemiological surveillance and policy. Phylogeographic methods to infer sampled viruses’ dispersal have been widely applied to quantify SARS-CoV-2 introductions into the UK (du Plessis et al., 2021), the USA (Deng et al., 2020; Gonzalez-Reiche et al., 2020; Worobey et al., 2020; Zeller et al., 2021), Brazil (Candido et al., 2020), New Zealand (NZ) (Douglas et al., 2021; Geoghegan et al., 2020), and Europe (Hodcroft et al., 2021; Worobey et al., 2020), among others, elucidating variable epidemic dynamics. A recent review expounds further on the approaches and application of phylodynamic models towards improving our understanding of SARS-CoV-2 transmission and control (Attwood et al., 2022). In the UK, where the most sequences have been generated per case globally (Furuse, 2021), there were an estimated 1179 introductions resulting in two or more sampled descendant cases in the early epidemic (du Plessis et al., 2021). By contrast, the first wave of COVID-19 in Louisiana, USA, was primarily attributable to a single domestic introduction several weeks prior to Mardi Gras, resulting in multiple superspreader events and wide dissemination across Southern USA (Zeller et al., 2021). In NZ, where stringent border closures and lockdown measures were enacted early on, there were estimated to have been 277 introductions up to 1 July 2020, among which 19% resulted in multiple downstream cases (Geoghegan et al., 2020). Large-scale SARS-CoV-2 genomic epidemiology analyses in Canada have thus far been limited to a study on the early epidemic in the province of Quebec in which they conservatively estimated at least 500 viral introductions into Quebec by June 2020 largely attributable to the province’s spring break (Murall et al., 2021). Our analyses elaborate upon their findings at a national scale for the first and second waves of COVID-19 in Canada. Characterization of viral importations over time can also clarify the effectiveness of public health interventions by associating inflection points in importations with drastic changes in policies such as international travel restrictions (Magalis et al., 2020). Compartmental modelling approaches have also been applied to quantify the impact of COVID-19 control measures such as social distancing, informing decisions about when to relax or increase stringency (Anderson et al., 2020; Anderson et al., 2021). Evaluating the effectiveness of these interventions is key to reacting proportionally to the risk posed by future outbreaks of SARS-CoV-2 and other zoonotic pathogens. It is unclear how changes in intervention stringency, social and travel behaviour, and circulating viral diversity affected Canadian SARS-CoV-2 transmission dynamics, particularly during a time of negligibly low immunity prior to vaccine roll-out or widespread natural infections. The first COVID-19 case in Canada was detected on 25 January 2020 in a traveller from Wuhan to Toronto and by 5 March, the first community transmission case was identified (Press, 2021; Figure 1C). In subsequent weeks, the stringency of Canadian interventions increased rapidly, summarized by the Oxford stringency index (Hale et al., 2021). On 14 March, a travel advisory warning against all non-essential travel outside Canada was issued; on 18 March, travel restrictions on the entry of all foreign nationals (except from the USA) were implemented; 21 March travel restrictions were extended to the USA; and 24 March, a mandatory 14-day self-isolation for those returning from international travel was implemented. The stringency index during the first wave increased most rapidly on 16 March 2020, reaching the highest stringency between 1 and 3 April 2020. The first wave reached a maximum of 1108 average daily new COVID-19 diagnoses across Canada on 3 May 2020. While many of the earliest cases in Canada were attributable to A.1 and B.4 lineages in British Columbia (Figure 1—figure supplement 1), cases in March through June 2020 were primarily attributable to B.1* lineages and B.1.1* lineages in Ontario and Quebec, as well as lineage B.1.279 in Alberta in March and April. By June, in response to reductions in daily new diagnoses country-wide, restrictions slowly relaxed, resulting in a reduction of the stringency index, albeit with varying rates by province (Cameron-Blake et al., 2021). Fewer than 186 average daily new diagnoses occurred across Canada throughout summer 2020, mostly in Quebec, Ontario, Alberta, and British Columbia. Then, with the rising second wave in fall 2020, the stringency index rebounded in November 2020, although with simultaneous easing of entry exceptions for foreign nationals and quarantine shortening (Cameron-Blake et al., 2021). The second wave entered exponential growth in November 2020, as VOCs and VOIs began to displace wild-type lineages. Figure 1 with 2 supplements see all Download asset Open asset A timeline of the first and second waves of the Canadian COVID-19 epidemic up to 1 March 2021. (A) Rolling average daily new COVID-19 diagnoses in previous 7 days across Canadian provinces and territories, or the daily count where data was sparse prior to April 2020. (B) Rolling average daily clean SARS-CoV-2 sequences collected in previous 7 days, or daily sequences where data was sparse, uploaded to Global Initiative on Sharing All Influenza Data (GISAID) by 15 June 2021. Incomplete sample collection dates were inferred within time-scaled phylogenies. (C) The Oxford Stringency Index for Canada overlaid with key epidemiologic events and national-level public health restrictions. Figure 1—figure supplement 1 summarizes the frequencies of Pango lineages among daily Canadian sequences. Figure 1—figure supplement 2 compares monthly cases, sequences available, and sequences sampled globally and within Canada. The VOCs and VOIs first detected in Canada were B.1.1.7 (Alpha) on 8 November 2020, followed by P.2 on 24 November, A.23.1 on 2 December, B.1.429 on 5 December, B.1.525 (Eta) on 14 December, B.1.351 (Beta) on 19 December, B.1.427 (Epsilon) on 30 December (dates from GISAID). The winter holidays were followed by the crest of the second wave on 11 January 2021 with 3555 average daily new diagnoses in Canada, likely dampened by restrictions on travel, personal gatherings, dining, and mask use. The new year also brought first Canadian detections of B.1.526.1/.2 (Iota) on 1 January 2021, P.1 (Gamma) on 1 February, and C.37 (Lambda) on 15 February. By March 2021, the trough of the second wave gave heed to the third wave, driven almost exclusively by B.1.1.7 and P.1, the latter in Western provinces primarily. Monitoring lineage frequencies alone does not inform how many individual importations accounted for the detected cases, nor the domestic spread dynamics of individual sublineages, warranting a genomic epidemiology analysis. A phylogenetic maximum likelihood approach was applied to estimate the timing, origins, destinations, and transmission of SARS-CoV-2 introductions in Canada from the beginning of 2020 to the end of the second wave in March 2021 for Canadian sublineages (introductions with sampled descendants) and singletons (introductions with no sampled descendants). We tested the hypothesis that international travel restrictions enacted in March 2020 effectively reduced international importations of SARS-CoV-2 into Canada, yet ongoing introductions contributed to COVID-19 persistence into early 2021, exacerbated by highly transmissible B.1.1.7 and other VOC sublineages. These analyses help to elucidate the relative contributions made by sublineages introduced before and after enactment of travel restrictions towards persistence of the Canadian SARS-CoV-2 epidemic in 2020 and early 2021, prior to the global predominance of VOCs and vaccine roll-out. Evaluating travel restrictions’ effectiveness towards reducing importations could inform the stringency and timing of future public health interventions. Results Global SARS-CoV-2 phylogeny with a Canadian focus The overrepresentation of sequences from countries with the highest sequencing efforts, such as the UK, was reduced in the dataset by subsampling (Figure 1—figure supplement 2; Figure 6—figure supplement 2). Cumulative sequence representation was increasingly comparable across provinces with fewer Canadian sequences subsampled, however excluding too many sequences resulted in missed introductions. There were no sequences available from the Yukon, Northwest Territories, Nunavut, or Prince Edward Island, which cumulatively had 603 COVID-19 cases (0.07% of Canada cases) prior to 1 March 2021. Canadian viral genomes were dispersed throughout the global phylogeny and represented 345 unique Pango lineages (Figure 2). The average time to the most recent common ancestor (TMRCA) was estimated as 22 December 2019 (24 November–23 December) across 10 subsamples, consistent with the upper end of others’ credibility intervals (Lu et al., 2020; Worobey et al., 2020). The relaxed molecular clock model estimated in LSD2 had a mean rate of 2.86 × 10–4 (2.84 × 10–4–2.92 × 10–4) substitutions/site/year (s/s/y), lower than the mean strict clock rate inferred in TempEst of 7.15 × 10–4 (6.83 × 10–4–7.48 × 10–4) s/s/y (Figure 2—figure supplement 2) due to differing assumptions of rate variation. Increasing the representation of global samples relative to Canadian sequences generally resulted in earlier TMRCA estimates, higher estimated clock rates (sampling wider diversity), more temporal outliers removed, and lower R2 values associated with their clock models (Supplementary file 3c; Supplementary file 3d). The large majority of sublineages (75.1%) were supported by high likelihoods (>0.9) in both introduction and parental nodes across all bootstraps (Figure 2—figure supplement 3), consistent across subsampling strategies. The 14 sublineages with 500 or more Canadian descendants cumulatively accounted for 59.1% of Canadian sublineage descendants, despite representing only 2.3% of sublineages (Figure 2). Figure 2 with 3 supplements see all Download asset Open asset Key Canadian sublineages in a phylogenetic tree of SARS-CoV-2 in Canada and globally up to the end of the second wave on 28 February 2021. (A) The highest likelihood bootstrap time-scaled phylogenetic tree inferred using a subsampling strategy where 75% of available Canadian sequences were retained and the remainder of sequences up to 50,000 were from global sources. (B) The timing and expansion of key Canadian SARS-CoV-2 sublineages with more than 500 sampled Canadian descendants. Height reflects the relative density of sampled Canadian descendants within each sublineage. Diamonds and dashed lines show the mean and 95% confidence interval of the time to the most recent common ancestor (TMRCA). Figure 2—figure supplement 1 zooms in on the subtrees for the four largest Canadian sublineages. Figure 2—figure supplement 2 summarizes the process of removing temporal outliers to improve molecular clock signal. Figure 2—figure supplement 3 summarizes sublineage introduction node and parent node likelihoods overlaid for all bootstraps. Diverse origins of Canadian SARS-CoV-2 sublineages Maximum likelihood discrete ancestral state reconstruction of the time-scaled phylogenies with 75% of Canadian sequences yielded a mean of 680 (658–703) Canadian sublineages (introductions with sampled descendants) by the end of February 2021 (Figure 3) and 1582 (1501–1663) singletons (introductions with no sampled descendants) (Figure 3—figure supplement 2). Although singletons accounted for 70% (69–71%) of introductions, some were likely representative of sublineages with no other sampled descendants. The proportion of importations resulting in domestic transmission did not vary significantly by province, origin, or month (Kruskal-Wallis test: p=0.53, p=0.47, p=0.37), but notably reached a maximum of 0.94 in February 2020 (Figure 6—figure supplement 3). Figure 3 with 5 supplements see all Download asset Open asset Flow of SARS-CoV-2 sublineages into Canadian provinces from global origins coloured by Pango lineage in (A) the first wave, before 1 August 2020, and (B) the second wave, from 1 August 2020 to 28 February 2021. Locations’ relative size and lineage flows between location pairs represent the mean percent of sublineages across 10 subsamples for the 75% sampling strategy. Lineages by location pair associated with greater than 0.5% of sublineages and locations associated with more than 1% of sublineages were labelled. Sublineages’ relative sizes, based on the number of sampled unique Canadian descendants during the (C) first and (D) second wave. Canadian sublineages were named based on the predominant Pango lineage of descendants and a ‘can’ suffix with a numeric denoting their order of first Canadian sample date. Figure 3—figure supplement 1 shows the flow of singletons. Figure 3—figure supplement 2 summarizes the distribution of sublineage sizes. Figure 3—figure supplement 3 shows the distribution of all Canadian sublineages’ descendants’ sampling locations. Figure 3—figure supplement 4 stratifies second wave sublineage sizes by pre- and post-January 2021. Figure 3—figure supplement 5 stratifies sublineages’ overall size by global and Canadian descendants. We estimated that during the first wave, from 1 January to 31 July 2020, there were a total of 384 (368–400) introductions resulting in domestic transmission, including 169 (155–182) B.1 sublineages, 53 (50–56) B.1.1 sublineages, 23 (20–25) A.1 sublineages, 9 (8–10) B.4 sublineages, 8 (7–9) B.1.36 sublineages, and 7 (6–7) B.40 sublineages (Figure 3A). Most first wave Canadian sublineages originated from the USA (49%, 46–52%), followed by Russia (12%, 11–13%), Italy (6%, 5–9%), India (6%, 5–7%), Spain (5%, 4–5%), and the UK (4%, 4–5%), among others. These sublineages were inferred to have been primarily imported into Quebec (39%, 38–40%) followed by Ontario (36%, 35–38%), and British Columbia (12%, 11–13%), followed by Manitoba (5%, 5–5%), the Maritimes (5%, 5–5%), Alberta (2%, 2–2%), and Saskatchewan (1%, 0–1%). The second wave, from 1 August 2020 to 28 February 2021, included 296 (280–313) newly introduced sublineages, primarily of B.1.2 (55, 54–56), B.1.1.7 (30, 19–40), B.1.36 (24, 22-25), B.1 (14, 12–15), and B.1.1.519 (8, 8–9) lineages (Figure 3B). The origins of sublineages introduced during the second wave remained predominantly from the USA (43%, 40–45%), alongside increased relative contributions from India (16, 15–17%), the UK (7%, 5–9%), Asia (6%, 5–6%), Europe (4%, 4–5%), and Africa (3%, 3–4%). There were relatively more sublineages introduced to Ontario (45%, 44–47%), British Columbia (23%, 22–24%), and Alberta (11%, 10–11%) during the second wave, and fewer to Quebec (16%, 16–17%) than in the first wave. Relative contributions of origins and destinations were mostly robust to subsampling strategy, although differed in their absolute estimates (Figure 6—figure supplement 4). Sublineage sizes were overdispersed (Figure 3—figure supplement 2) with a mean of 53 (47–59) total unique sampled descendants (globally and within Canada) and a median of 4 (4–4). There were 46 (44–49) sublineages with more than 100 unique sampled descendants. Together, they represented 6.8% (6.4–7.2%) of sublineages and 0.7% (0.6–0.8%) of total introductions including singletons, yet their Canadian descendants accounted for 77% (74–81%) of Canadian sequences. Only 14 sublineages had 500 or more sampled Canadian descendants (Figure 2). The largest sublineages from the first wave, including B.1.279.can1, B.1.can1, B.1.128.can1, and B.1.1.176.can1, persisted through to the second wave (Figure 2—figure supplement 1), although their descendants no longer accounted for the majority of Canadian samples (Figure 3D). In the second wave, the most frequently sampled sublineages in Canada were AE.8.can1 (introduced in July 2020), B.1.36.18.can1 (July 2020), B.1.1.7.can1 (November 2020), and B.1.438.1.can1 (September 2020). As the second wave progressed into the new year, increasingly VOC sublineages, B.1.1.7.can2 and B.1.1.7.can3, outcompeted previously dominant sublineages, B.1.36.18.can1, B.1.128.can1, and AE.8.can1 (Figure 3—figure supplement 4). A total of 34,169 (31,337–37,001) sampled descendants of Canadian sublineages were identified (Figure 3—figure supplement 3). While the majority of descendants were in Canada (n=25,270, 25,071–26,369), the 8449 (5915–10,983) global descendants were vastly underestimated by subsampling. Travel restrictions reduced sublineage importations Sublineages’ TMRCA approximates the date of the first transmission event resulting in one or more sampled descendant following viral Although TMRCA is not to the date of a viral host a they are associated by a sampling and one or more By sublineages’ TMRCA over the temporal dynamics of SARS-CoV-2 importations were by global origin and Canadian (Figure 4). There have been of 14 sublineages introduced prior to the first COVID-19 case in Canada identified on 25 January 2020, and sublineages prior to the implementation of travel restrictions for all foreign nationals on 21 March 2020. early sublineage introduction dates have wide due to low phylogenetic diversity and large On the following of travel restrictions April 2020), the importation rate reached maximum of 58.5 (50.4–66.5) sublineages per including sublineages per week from the USA and sublineages per week introduced into Ontario had reached maximum several days before on 19 March, at sublineages per many were from the yet the mandatory 14-day quarantine was not enacted 25 By 5 2 weeks after travel restrictions the overall sublineage importation rate had and within 4 weeks, the rate had 10.3-fold This inferred inflection in sublineage importation rate was robust to multiple subsampling (Figure 6—figure supplement The to which importations were reduced within 2 weeks of stringency by importations were reduced by in the in Quebec, in Ontario, in British and in dynamics for singletons mostly sublineage (Figure supplement Figure 3—figure supplement 1), although the USA contributed relatively more towards singletons than sublineages in both and the maximum importation rate was higher in the first wave than the second wave. Figure 4 with 2 supplements see all Download asset Open asset introduction rates of Canadian SARS-CoV-2 sublineages in the first two in the of changes in COVID-19 intervention The sublineage introduction rates were summarized as across bootstraps (A) by global origin and (B) by province of The to of high and low stringency, based on Oxford COVID-19 Stringency Figure supplement 1 summarizes introduction Figure supplement 2 changes in sublineages’ and over Despite introductions were at a low sublineages per 1 August 2020, when a in importations was detected into the second wave. further of travel restrictions for international and in 2020 for 2021), with a sustained low stringency, importation rates rebounded to sublineages per week on 2 with increased travel prior to the rates increased to sublineages per week on 3 to the new year, VOC and sublineages had been introduced to Canada, including 15 5 2 5 and 4 A.23.1 sublineages. Sublineage and over time to elucidate the relative contributions of early and sublineages, sublineage size of sampled descendants) over time was (Figure supplement were by they were sampled Canadian cases in 2 before 1 March or to for in sublineages’ time to and sample Sublineage size was significantly reduced over time among and sublineages sublineages’ transmission from TMRCA to most recent Canadian over time for both and sublineages Figure supplement a hypothesis where both earlier and sublineages were likely to be and the mean number of days since importation to be over While the average number of days since importation of Canadian samples from sublineage TMRCA to sample increased over time during the first wave, several in summer 2020 as new sublineages more the since importation in the second wave (Figure supplement sublineage from TMRCA to first Canadian did not significantly over time Figure supplement There was that was significantly associated with province of introduction (Kruskal-Wallis and transmission The inferred transmission source likely state at node of all sampled Canadian genomes was investigated to the relative of domestic and international transmission by province over were as the USA, or other international sources. Following the entry for foreign nationals in there was a but not an of the proportion of transmission events attributable to the USA or other international across all provinces (Figure to the restrictions, provinces had a mean of of transmission events with international origins, from in Saskatchewan to in British Columbia. In April 2020, following travel restrictions, all provinces had of international transmission yet Ontario and Quebec retained relatively high mean number of transmission events attributable to international at and (Figure supplement This implementation or with quarantine in these Figure 5 with 1 supplement see all Download asset Open asset Relative contributions of international and domestic transmission sources. (A) and total contributions of the USA, other and transmission among all sampled by province and month between in March, and May 2020. (B) The flow of lineages among Canadian before and after August 2020. the mean across bootstraps for the 75% subsampling strategy. Figure supplement 1 the total number of sampled transmission events with international source across Canada in April 2020. transmission was investigated by transmission origins among Canadian sequences with an inferred Canadian origin from province (Figure the first wave, Ontario and Quebec were the sampled of transmission, with and for the largest In the second wave after August 2020, Ontario remained the of followed by nearly contributions from British Alberta, and lineages in the second wave were from Alberta to British and from Ontario to Quebec, from British Columbia to Quebec and Ontario to Quebec, and B.1.1.7 from Ontario to phylogenetic reduce overrepresentation of with more
Source: eLife
Author: Sarah E Cobey
Institution: University of Chicago
Language: English
Subfield: Infectious Diseases
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,500
OpenAlex ID: W4289539254