Association of E484K spike protein mutation with SARS-CoV-2 infection in vaccinated persons---Maryland, January – May 2021

Abstract Among 9,048 people infected with SARS-CoV-2 between January-May, 2021 in Maryland, in regression-adjusted analysis, SARS-CoV-2 viruses carrying the spike protein mutation E484K were disproportionately prevalent among persons infected after full vaccination against COVID-19 as compared to infected persons who were not fully vaccinated (aOR 1.96, 95% CI, 1.36 to 2.83).

This retrospective study -conducted while the prevalence of lineage B.1.617.2 "Delta" in Maryland was less than 1% -examines whether the E484K and L452R substitutions were associated with infection in vaccinated persons using genetic sequences from SARS-CoV-2 specimens collected in the U.S. State of Maryland.

Methods
This retrospective analysis uses all specimens on GISAID's public online repository of SARS-CoV-2 sequences collected between the start of January 2021 and the second week of May of 2021 that were collected from Maryland residents and could be successfully linked to a case in Maryland  Hospitalization of a person with full or any vaccination was defined the same, with the added condition that the patient was admitted to the hospital within 28 days following collection of their first positive specimen.
Using logistic regression, we estimated the respective crude and adjusted associations of of E484K and L452R being present in a sequence with the corresponding SARS-CoV-2 specimen being collected from a vaccinated person (as defined above). Logistic regression was used to adjust for confounding by the following characteristics: age group, gender, region of residence, laboratory submitting to GISAID, and week of specimen collection.
The analysis was repeated for each of three vaccine types separately. For vaccine-specific analysis, persons vaccinated with a vaccine other than the type examined in that analysis were excluded (e.g., persons vaccinated with mRNA-1273 or Ad26.COV2.S vaccines were excluded from analysis of the BNT162b2 vaccine).
Regression coefficients were exponentiated, and interpreted as odds ratios. Wald confidence intervals (95%) were estimated for all regression coefficients; odds ratios with confidence intervals not containing 1 are statistically significant at the p<0.05 level.

Three sensitivity analyses are described in online supplemental Appendix A.
A c c e p t e d M a n u s c r i p t 5

Results
Of 9,048 SARS-CoV-2 specimens on GISAID collected from Maryland residents during the study period and included in this analysis, 1,187 (13.1%) carried the E484K substitution and 731  A c c e p t e d M a n u s c r i p t 6 or from persons with any vaccination (OR 1.07, 95% CI, 0.54 to 2.12; aOR 0.85, 95% CI, 0.41 to 1.78) (Table 1).
In adjusted analysis, for both E484K and L452R, associations with infection in fully vaccinated persons were comparable for all vaccine types (Table 1).

Discussion
In Maryland, in the first half of 2021, infections in vaccinated persons who received some or all scheduled doses of a COVID-19 vaccine were uncommon. When they did occur, infections in vaccinated persons were disproportionately likely to be viruses that carried E484K. This is consistent with past research that hypothesized E484K reduces the effectiveness of acquired immunity more than other mutations, possibly in part because that mutation is located in core RBD epitope of the spike protein; by contrast L452R is in a loop epitope [2,10]. Importantly, while hospitalizations following SARS-CoV-2 infection in vaccinated persons were rare in this study, we find that the association of E484K with infection in vaccinated persons persisted even if we restricted the outcome only to persons who were hospitalized. Additional studies focused on severe illness are needed.
We did not find an association of L452R with infections in vaccinated persons. This is in contrast to in vitro studies showing L452R is associated with reduced neutralization from polyclonal antibodies, albeit less strongly than E484K [10], and to epidemiologic evidence of reduced vaccine efficacy for lineage B.1.617.2 "Delta", which carries L452R [11]. B.1.617.2 was less than 1 percent of all sequenced specimens in Maryland during this study period. Past studies examined lineages, rather than specific mutations. If L452R was one of several mutations in those lineages contributing to reduced neutralization, its individual contribution might be too small to detect in this study.
During the period of this study, the rate of confirmed COVID-19 cases in Maryland fell 10fold [12]. At the same time, the fully vaccinated increased from fewer than 1,000 to approximately A c c e p t e d M a n u s c r i p t  In., and the decision was made that this project was nonresearch and did not require M a n u s c r i p t 11 Tables   Table 1. Crude and adjusted odds ratios for association of E484K and L452R mutations with post-vaccination SARS-CoV-2 infection, by hospitalization status and vaccine type