Higher prevalence of B.1.1.519 in Alaska versus the contiguous United States
We found a striking difference in the percentage of sequenced cases, a metric that can be used as an estimate of prevalence, assigned to the lineage B.1.1.519 between Alaska (Fig. 1B) and the rest of the United States (Fig. 1D) in early 2021. While the earliest detection of B.1.1.519 in the United States occurred 27 December 2020 in sequence data from New York, it was not until five weeks later, 4 February 2021, that B.1.1.519 was detected in sequence data from Alaska. By this time in the United States, other VOCs including Alpha (B.1.1.7) and Epsilon (B.1.427/429) had already emerged and were being consistently detected in sequence data. In contrast, Alaska had detected a single case of Alpha and no cases of Epsilon in December of 2020.
Unlike the contiguous United States where Alpha comprised approximately 5% of sequenced cases by the end of January, in Alaska, Alpha was not consistently detected in sequenced cases until March of 2021 (Fig. 1B). By the first week of March in Alaska, B.1.1.519 had already reached an estimated 60.8% of sequenced cases compared to 4.6% in the rest of the US, which was the peak prevalence for the contiguous United States (Fig. 1B). During B.1.1.519’s peak in the contiguous United States, Alpha made up 30.5% of sequenced cases. By contrast, in Alaska, B.1.1.519 reached a peak prevalence of 77.9% by the week of 4 April 2021, 5 weeks after the peak in the contiguous United States.
In terms of COVID-19 cases reported in Alaska, there was a sharp decline the first several weeks of December 2020 followed by an increase in cases through the first week of January 2021 and then another decline (Fig. 1A). During this time, cases in the United States increased until the second week of January 2021 followed by a decline through the third week of February 2021 (Fig. 1C). In both Alaska and the rest of the United States, new COVID-19 cases remained at a relatively stable rate while the cases attributed to VOC, VOI, and VBMs were rising (Fig. 1A,C). Given the complexity of factors that affect case rates, it is difficult to attribute the stabilization in cases to biological mechanisms driven by variants or behavioral and social dynamics such as increased vaccination rates, social restrictions, or adherence to respiratory hygiene measures during this time.
When B.1.1.519 first emerged in Alaska, several restrictions were in place as a result of COVID-19 Public Health Disaster Emergency Declaration enacted 11 December 2020 and then renewed 14 January 2021 (Fig. 1A). The State of Alaska recommended wearing a face covering in most situations, working remotely when possible, and avoiding gatherings with non-household members. In December, Alaska also received its initial allocations of the Comirnaty vaccine (Pfizer-BioNTech BNT162b2) for frontline workers. By 11 January 2021, Alaskans 65 years of age and older were eligible for both Comirnaty and Spikevax (Moderna mRNA-1273). Vaccine availability was expanded to the entire public by 10 March 2021 and included primarily Spikevax and Comirnaty, at which time B.1.1.519 comprised 60.8% of the sequenced cases. Alaska was ahead of the contiguous United States in terms of vaccine availability and percent of the population vaccinated in early 2021, which confounds the role host immunity may have had in variant emergence and infection dynamics. Citing the the rapid vaccination rates, Alaska ended the COVID-19 Disaster Declaration, signally and end to mitigation measures on April 30, 2021.
Difference in the emergence of B.1.1.519 within Alaska’s two most populated regions
Given Alaska’s vast geographic expanse and limited travel options, populations of people tend to interact more frequently within economic regions of the state. In Alaska there are six economic regions defined by the Department of Labor and Workforce Development: the Anchorage-Mat Su, Interior, Gulf Coast, Southeast, Southwest, and Northern regions in order from highest to lowest population. Based on available but limited data, we observed distinct trends in SARS-CoV-2 across economic regions of Alaska (Fig. S1). However, in our analysis, we excluded all but the two most populous economic regions, the Anchorage-Mat Su and Interior, because of the low proportion of sequenced cases in the other economic regions of the state. We can draw more robust conclusions about estimates of prevalence because these two regions sequenced greater than 5% of the newly reported cases February through June of 2021 (Fig. S2).
Between the Anchorage-Mat Su and Interior economic regions there were distinct dynamics in the prevalence of variants over time (Fig. 2). For example, Alpha was consistently detected earlier in the Anchorage-Mat Su (Fig. 2B) than the Interior (Fig. 2D), 360 miles apart. However, in the Interior, Epsilon comprised a high percentage of sequenced cases from January through March 2021 (Fig. 2D), and the emergence and persistence of B.1.1.519 differed between these two regions of the state.
The first case of B.1.1.519 in Alaska was detected from a sample collected 4 February 2021 in the Southwest region of the state. In both the Anchorage-Mat Su and Interior regions, B.1.1.519 was detected in specimens with collection dates one day apart from one another, 7 and 8 February 2021, respectively. By the end of February, B.1.1.519 comprised 37.8% of sequenced cases in the Anchorage-Mat Su region and 10.5% of cases in the Interior (Fig. 2). By February VBMs detected in the Anchorage-Mat Su included Alpha, Epsilon, and Gamma. On the other hand, in the Interior, Epsilon was the only VBM detected and comprised 38.5% of sequenced cases in February (Fig. 2D). In the Interior, it appears that Epsilon, which was first detected 6 January 2021, was able to establish itself before B.1.1.519. However, one month after B.1.1.519 was first detected in the Interior, B.1.1.519 overtook Epsilon. By April, the majority of sequenced cases were assigned to B.1.1.519 in the Interior (77.2%) and the Anchorage-Mat Su (72.6%). In both regions, there is a noticible increase in cases around this peak in B.1.1.519 prevlaence (Fig. 2A,C). After peaking in April, prevalence of B.1.1.519 in May only marginally decreased for the Interior to 74.1% but sharply decreased for the Anchorage-Mat Su region to 38.8% (Fig. 2).
The sharp decline of B.1.1.519 in the Anchorage-Mat Su region may have occurred earlier than in the Interior due to the emergence of other variants that had established themselves in the Anchorage-Mat Su region well before the Interior. For example, in the Anchorage-Mat Su region, Alpha was detected on 20 December 2020—97 days before it was first detected in the Interior. This gave Alpha more time to establish itself within the population and consequently reached 42.9% of sequenced cases by May in the Anchorage-Mat Su region versus 9.8% in the Interior. Delta also emerged in the Anchorage-Mat Su region weeks before it was first detected in the Interior, highlighting why B.1.1.519 may have persisted in Interior populations with the last case of B.1.1.519 detected 2 weeks after last detection in the Anchorage-Mat Su.
High prevalence of B.1.1.519 in Alaska driven mutational advantage
Mutations in the SARS-CoV-2 genomes from Alaskan cases occurred in an almost clockwise fashion (Fig. 3A). Although most of these mutations are expected to have neutral effects, some changes can confer selective advantages that enhance fitness by increasing immune evasion, transmissibility, and/or infectivity16. Mutations that cause amino acid changes to the spike (S) glycoprotein may confer selective advantages given the role of the S protein in COVID-19 pathogenesis and tropism. The mature S protein is cleaved into a fusion domain (S2), and a S1 domain containing an N-terminal head and the receptor binding domain (RBD) that binds to host cell human angiotensin converting enzyme-2 (hACE2) receptor. The S1 domain and RBD particularly are key targets for binding of neutralizing antibodies induced by infection or vaccines, principally IgG17. Given the role of the S protein in receptor binding and subsequent membrane fusion and viral entry into host cells, mutations affecting S1, especially the RBD, can impact hACE2 affinity, viral entry, infectivity, and immune evasion18,19.
B.1.1.519 has 11 shared amino acid (AA) mutations relative to the original SARS-CoV-2 (Wuhan-1) genome, with 4 in S: T478K and D614G in the RBD, P681H in the S1/S2 cleavage site, and T732A in the S2 domain5. T478K also arose in B.1.617.2 (Delta)20 and may contribute to antibody evasion21. P681H also occurs in B.1.1.7 (Alpha) and has a weak ability to increase proteolytic cleavage of S1/S2 in vitro22. The functional significance of B.1.1.519 harboring these mutations in S is unclear. In a study from the Netherlands, there was no evidence of increased viral load conferred by B.1.1.51923.
Interestingly, based on the regression of total substitutions over time, which exhibited an increase at a steady rate, B.1.1.519 genomes had more total substitutions across the genome (Fig. 3A, red line) than the background of non-emerging lineages (Fig. 3A, gray and black lines). B.1.1.519 had significantly more AA substitutions in the spike protein than lineages detected in Alaska prior to the week of B.1.1.519’s peak prevalence, 4 April 2021, suggesting a potential competitive advantage over previously circulating lineages (B.1.1.519 = 4.18 ± 0.01, Pre-peak lineages = 2.65 ± 0.10 S AA substitutions, Wilcoxon W = 42,302, p-value < 2e−16; Fig. 3B). The period of time after B.1.1.519’s peak prevalence was dominated by the emergence of many VOC. Genomes sequenced after B.1.1.519’s peak prevalence had significantly more AA changes in the spike protein than B.1.1.519, dominated by more divergent Alpha (Fig. 3A, yellow line) and Delta VOC (Post-peak lineages = 7.75 ± 0.07 S AA substitutions, Wilcoxon W = 1e+05, p-value < 2e−16; Fig. 3B). These VOC likely held a selective advantage that allowed them to outcompete B.1.1.519 in Alaska, a phenomenon of variant replacement observed across Alpha and Delta waves in the USA24.
B.1.1.519 established itself within the circulating population of SARS-CoV-2 viruses in Alaska weeks before other variants, such as Alpha, were consistently detected (Fig. 1B). The difference in prevalence of B.1.1.519 during Alpha’s emergence in Alaska versus the contiguous United States, paired with the finding of more S protein amino acid changes in this variant than previous lineages in Alaska, suggests that in Alaska B.1.1.519 likely emerged and became dominant due the unique diversity of variants. In this context, the initial transmission events of B.1.1.519 led to this variant outcompeting the previously circulating lineages because of a selective advantage that that variant already possessed. The same trend was likely not possible in the other states because Alpha, which proved to outcompete B.1.1.519, had already established itself within the population, outcompeting B.1.1.519 when it emerged.
Other locations displayed similar patterns to Alaska in terms of B.1.1.519’s emergence and spread, most notably Mexico. B.1.1.519 was first detected in Mexico in November 2020, about three months prior to Alaska’s first detection, and was found to rapidly outcompete existing variants to become the dominant variant in the country, comprising 51.5% of sequenced genomes by January 20215. By February, 90.9% of Mexico City’s sequenced cases were B.1.1.519. In Mexico, B.1.1.519 was not displaced by ongoing transmission of Alpha25. The pattern of circulation and displacement is similar to Alaska. In the comprehensive survey of variants in Mexico, Taboada et al. attributed the success of B.1.1.519 to the unique combination of circulating lineages at the time of arrival26.
In a study examining the clinical impacts of B.1.1.519 in Mexico City, Cedro-Tanda (2021) found that B.1.1.519-infected patients had a significant increase in adjusted odds ratio for developing severe symptoms of COVID-19 including dyspnea, chest pain and cyanosis, and hospitalizations compared to non-B.1.1.519 infected patients. Unfortunately, the necessary data to draw conclusions about the clinical impacts of B.1.1.519 in Alaska, as shown in the Mexico-City study, were not available. Another study, from the Netherlands, describes a cluster of cases associated with B.1.1.519, demonstrating a high within-facility attack rate23. Interestingly, that same study did not find differences in B.1.1.519 viral loads compared with other variants, and sample size limited conclusions about disease severity.