Genomic characterization and epidemiology of an emerging SARS-CoV-2 variant in Delhi, India

Abstract

Delhi, the national capital of India, experienced multiple severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) outbreaks in 2020 and reached population seropositivity of >50% by 2021. During April 2021, the city became overwhelmed by COVID-19 cases and fatalities, as a new variant, B.1.617.2 (Delta), replaced B.1.1.7 (Alpha). A Bayesian model explains the growth advantage of Delta through a combination of increased transmissibility and reduced sensitivity to immune responses generated against earlier variants (median estimates: 1.5-fold greater transmissibility and 20% reduction in sensitivity). Seropositivity of an employee and family cohort increased from 42% to 87.5% between March and July 2021, with 27% reinfections, as judged by increased antibody concentration after a previous decline. The likely high transmissibility and partial evasion of immunity by the Delta variant contributed to an overwhelming surge in Delhi.

Fig. 1.. Multiple surges of SARS-CoV-2 infections in Delhi with an overwhelming outbreak in April and May 2021.

(A) Weekly tests, confirmed cases, and test positivity rate in Delhi from April 2020 to June 2021. Sample collection period for CSIR serosurveys is marked as P1 to P3. (B) Number of hospitalizations and ICU admissions plotted on a daily basis from June 2020 to June 2021. The arrow marks the possible saturation of ICU capacity (3). (C) Daily cases and daily deaths from January 2021 to June 2021. (D) Time-advanced and scaled cumulative cases, fitted to cumulative deaths. Time advancement of cumulative reported cases by 8 days was done for maximal coincidence with scaled cumulative deaths. Case fatality ratio = averaged scaling factor (cumulative deaths / time-advanced cumulative cases). Mean ± SD, 0.019 ± 0.003.

Fig. 2.. Serological estimates of prior infections, preexisting immunity, and new infections for the April and May 2021 outbreak.

(A) Seropositivity in the CSIR cohort, subdivided by nature of employment and use of public transport, plotted for different time periods (phase I to phase III, proportion ± 95% CI). Details are provided in table S1. (B) Variability and temporal decline in neutralization capacity estimated by sVNT assay between phases I and II (n = 52 subjects). (C) Serial antibody concentration measurements in initially seropositive subjects (n = 91). Pattern suggestive of reinfections is shown (decline followed by rise, n = 25). (D) Remaining data (n = 66 subjects), with four indeterminate reinfection cases indicated with arrows. Antibody concentration is reported in multiples of the assay cutoff index value (CoI).

Fig. 3.. Genomic-epidemiologic correlations.

(A) Time trends of Ct values (mean ± SE) and high viral load samples (proportion ± SE) for Orf1 gene (E gene data, fig. S1). (B) Smoothed graph of main lineages in Delhi from March 2020 to May 2021 in biweekly increments. New cases and TPR are aligned and plotted on the same timeline. (C) Phylogenetic analysis for variant of concern (VOC) strains between Delhi and states (Punjab and Maharashtra) with known VOC outbreaks before April 2021. Further analysis suggesting a superspreading event for Alpha is shown in fig. S3. (D to F) Month-wise proportions of different lineages (n > 3 sequenced samples) in states surrounding Delhi. Additional data are shown in figs. S4 and S5.

Fig. 4.. Estimates of the epidemiological characteristics of the Delta variant.

Values were inferred from a two-category Bayesian transmission model fitted to mortality, serosurvey, and genomic data from Delhi, India. (A) Joint posterior distribution, with isoclines corresponding to the 90% and 50% enclosures of posterior density of the Delta variant immune escape and transmissibility increase relative to non-Delta categories. Immune escape has a median of 20% with 50% Bayesian credible interval (bCI) of 10 to 50%, and transmissibility has a median increase of 1.5 with 50% bCI of 1.3 to 1.7. (B) Delta fraction over time, inferred by the model. Black dots represent genome sampling data points, with exact binomial confidence intervals. (C) Serosurvey data (black dots) and inferred cumulative incidence for Delta and non-Delta variant categories. (D) Mortality data (black dots) and inferred deaths assuming 50% underreporting. Other underascertainment scenarios are presented in the supplementary materials.