The analysis used sera collected from the Household Influenza Vaccine Evaluation (HIVE) cohort between July 1 and December 31, 2021, representing a pre-Omicron interval. The cohort contributes repeated measures and active acute respiratory illness surveillance, which supplied detailed vaccination and infection histories for participants. Investigators selected 156 serum samples from 117 participants for this nested serosurvey. Most samples came from adults (65.8%); children under 5 years were underrepresented. Participants were classified into four pre-Omicron immunity groups: vaccinated only, prior infection only, hybrid immunity, or neither vaccinated nor infected. The majority in this sample received COVID-19 vaccination during the study window. Demographic details included a predominantly White, non-Hispanic sample and a notable proportion reporting at least one high-risk medical condition.
Investigators ran a head-to-head comparison of three assay approaches: multiplex electrochemiluminescence (ECL) for binding IgG, multiplex ACE2 inhibition to assess blocking of receptor binding, and pseudoviral neutralization (PN) for functional neutralizing titers. Within the ECL panels, Spike and RBD IgG correlations were very high (Spearman coefficients 0.954–0.996), indicating tight internal consistency across strains for binding measurements. Cross-platform correlations were also substantial. ECL IgG versus PN showed moderate-to-strong correlation (0.811–0.827). ACE2 inhibition versus PN correlations were higher (0.877–0.953) for matched variants, reflecting concordance between functional ACE2-blocking measures and neutralization. The authors reported that ECL IgG concentrations tended to be larger for the Beta (B.1.351) and Delta (B.1.617.2) antigens at comparable PN titers, a pattern observed in these matched comparisons.
Across 156 samples, 17 (10.9%) were nucleocapsid (N)-positive by the manufacturer cutoff. Of those N-positives, some had documented prior PCR-confirmed infection while others did not, reflecting imperfect capture by PCR-based surveillance or potential asymptomatic infections. Correlations among antigen pairs were high for Spike IgG (Spearman 0.991–0.999) and for RBD IgG (0.974–0.999). ACE2 inhibition correlations were high broadly but were lower when BA.2 antigens were compared to other variants; Spike ACE2 correlations for BA.2 versus other variants ranged from 0.436–0.714, and RBD ACE2 correlations from 0.329–0.435. These lower correlations suggest some antigenic divergence in functional blocking assays for BA.2-related targets relative to earlier strains.
Receiver operating characteristic analysis assessed how well each antigen and assay discriminated vaccinated (including hybrid) versus unvaccinated participants. ECL IgG antigens yielded consistent discrimination with area under the curve (AUC) values between 0.84 and 0.88. Optimal thresholds derived by Youden’s Index produced high sensitivity (0.88–0.91) and specificity (0.87–0.88) for IgG antigens. ACE2 inhibition assays generally demonstrated higher specificity (0.91–1.00) but variable sensitivity and AUC across variants. For Spike ACE2, AUC ranged from 0.64 for BA.2 to 0.96 for B.1.351, while RBD ACE2 AUC ranged from 0.53 (BA.2.75) up to 0.95 for Alpha and Delta antigens. Sensitivity estimates for ACE2 inhibition assays were heterogeneous (0.17–0.93) and lowest for several BA.2-related antigens, indicating reduced performance for those targets in classifying prior vaccination by ACE2 blocking metrics.
The study team modeled antibody concentrations by age using generalized additive mixed models. ECL IgG concentrations generally rose with age and appeared to plateau in early adulthood across the antigens measured. The pattern was consistent across Spike and RBD targets, driven by the cohort’s vaccination and infection histories captured through active surveillance. The authors cautioned that estimates among younger children were imprecise due to small sample sizes in those age strata. They also noted that individual-level data cannot be publicly posted because of identifying information tied to an ongoing longitudinal cohort; de-identified data are available to approved researchers on request per institutional review board restrictions. Additional limitations described by the study included the modest sample size overall and the underrepresentation of certain demographic groups, which may constrain generalizability.
Data processing and assay quality control followed manufacturer protocols and standardized R scripts; samples were run in duplicate with reflex to higher dilutions and multiple runs used to calculate coefficients of variation. The authors made their analysis code publicly available on GitHub. Remaining questions in the dataset include the extent to which ACE2 inhibition declines for later Omicron sublineages and how pediatric immune profiles would appear with larger, age-stratified sampling—issues the available data could not resolve given the study period and sample composition.