The use of representative community samples to assess SARS-CoV-2 lineage competition: Alpha outcompetes Beta and wild-type in England from January to March 2021

Abstract

Genomic surveillance for SARS-CoV-2 lineages informs our understanding of possible future changes in transmissibility and vaccine efficacy and will be a high priority for public health for the foreseeable future. However, small changes in the frequency of one lineage over another are often difficult to interpret because surveillance samples are obtained using a variety of methods all of which are known to contain biases. As a case study, using an approach which is largely free of biases, we here describe lineage dynamics and phylogenetic relationships of the Alpha and Beta variant in England during the first 3 months of 2021 using sequences obtained from a random community sample who provided a throat and nose swab for rt-PCR as part of the REal-time Assessment of Community Transmission-1 (REACT-1) study. Overall, diversity decreased during the first quarter of 2021, with the Alpha variant (first identified in Kent) becoming predominant, driven by a reproduction number 0.3 higher than for the prior wild-type. During January, positive samples were more likely to be Alpha in those aged 18 to 54 years old. Although individuals infected with the Alpha variant were no more likely to report one or more classic COVID-19 symptoms compared to those infected with wild-type, they were more likely to be antibody-positive 6 weeks after infection. Further, viral load was higher in those infected with the Alpha variant as measured by cycle threshold (Ct) values. The presence of infections with non-imported Beta variant (first identified in South Africa) during January, but not during February or March, suggests initial establishment in the community followed by fade-out. However, this occurred during a period of stringent social distancing. These results highlight how sequence data from representative community surveys such as REACT-1 can augment routine genomic surveillance during periods of lineage diversity.

Fig. 1.

The Alpha variant in England from January to March 2021. (a) Proportion of the Alpha variant over time. Points show raw data with error bars representing the 95% confidence interval. Shaded region shows the best fit Bayesian logistic regression model with 95% credible interval. (b) Odds ratio of a determined lineage being Alpha by age group for logistic models including just age group (purple) and both age group and region (green) fit to data from round 8 only. (c) Proportion of positive tests that are from the Alpha variant by region of England. Error bars show the 95% confidence intervals. (d) Proportion of positive tests that are from the Alpha variant by age group. Error bars show the 95% confidence intervals.

Fig. 2.

Geospatial patterns of lineage frequency. (a) Location of all positive samples for which we have identified their lineage for each round (each point moved randomly a small distance). (b) Modelled proportion of the Alpha variant across space for round 8, round 9, and round 10. Regions: NE = North East, NW = North West, YH = Yorkshire and The Humber, EM = East Midlands, WM = West Midlands, EE = East of England, L = London, SE = South East, SW =South West.

Fig. 3.

Symptoms and antibody positivity. (a) Proportion of those infected testing positive for antibodies 6 weeks after swab test (for all samples and for those that had both N- and E-gene detected), displaying any symptoms in the week prior to their swab test, displaying classic COVID-19 symptoms (loss of sense of taste, loss of sense of smell, new persistent cough, fever) in the week prior to their test and displaying no symptoms. (b) Odds ratios of the covariates of multiple logistic regression models. Each model had the result of the LFIA antibody test as the outcome variable with different combinations of lineage, N-gene Ct, E-gene Ct and age as the covariates. OR displayed for Alpha is relative to wild-type. OR displayed for N- and E-gene Ct is relative to a change in Ct of +5. OR displayed for age is relative to a change of +10 years in age.

Fig. 4.

Phylogenetic tree showing the relation of Beta lineages detected in REACT-1 to other Beta sequences in the COG-UK database. Sequences are coloured by the location in which the sequence was isolated. REACT lineages are coloured red and have an ID beginning with the sequence ‘ARCH-”- next to them. (a) The subgroup of the entire constructed tree that contains all REACT sequences, re-rooted to the COG-UK sequence SouthAfrica/KRISP−K006830/2020. (b–d) Zoomed-in view of the subtrees shown by the three shaded regions. Note that adjacent sequences ARCH-000047A3 and ARCH-000052A7 are multiple readings from the same individual.

Fig. 5.

Phylogenetic tree showing the relation of A.23.1 lineages detected in REACT-1 to other A.23.1 sequences in the COG-UK database. REACT-1 sequences are coloured in red and have an ID beginning with the sequence ‘ARCH-’ next to them. All other sequences are coloured by the location in which the sequence was isolated.

Fig. 6.

Patterns of frequency of returning from abroad in the prior 2 weeks. (a) Proportion of participants who answered they had been abroad in the previous 2 weeks by lower tier local authority. Regions: NE = North East, NW = North West, YH = Yorkshire and The Humber, EM = East Midlands, WM = West Midlands, EE = East of England, L = London, SE = South East, SW = South West. (b) Proportion of individuals who answered that they had been abroad in the previous 2 weeks by region and round. Dates: round 5=18 September–5 October 2020, round 6=16 October–2 November 2020, round 7=13 November–3 December 2020, round 8=6 January–22 January 2021, round 9=4 February–23 February 2021, round 10=11 March–30 March 2021.