In this carefully monitored cohort of household participants in Managua, Nicaragua, we found that both homotypic Omicron-neutralizing antibody titers (e.g., anti-BA.1 titer for BA.1 wave participants) and ancestral SARS-CoV-2-neutralizing antibody titers were informative correlates of protection against symptomatic infection during Omicron BA.1 and BA.2 infection waves in 2022. Mediation analysis further revealed that neutralizing antibodies causally mediate protection from infection conferred by vaccination or prior infection. Our findings confirm that neutralizing antibody titers serve as a meaningful correlate of protection, measuring a causally protective factor, that may inform risk assessment.

Many prior studies have demonstrated, to varying degrees, that humoral immunity serves as a correlate of protection from SARS-CoV-2 infection or severe symptoms^{2,3,4,5,6,7,9,15,17,18,19,20,21,22,23,24}. The results presented here align with previous findings on ancestral SARS-CoV-2-neutralizing titers but also extend beyond prior work in key ways. First, the participants in this study reside in Managua, Nicaragua, where SARS-CoV-2 had circulated extensively before the Omicron waves. The original SARS-CoV-2 strain started circulation in 2020, by 2021 Gamma and Delta predominated, and starting 2022 Omicron took over, first BA.1 and then BA.2 quickly replaced BA.1²⁵. By the time of this study’s sampling, people had been infected multiple times and subsequently vaccinated¹⁶. The vaccines administered in this population included Sputnik and AstraZeneca for adults, and Abdala and Soberana for children²⁶.

To the best of our knowledge, no prior studies on neutralizing antibody correlates of protection from COVID-19 have focused on participants in Nicaragua or any other Central American countries. One prior study from South America, which included participants from Chile and Peru, analyzed ancestral-neutralizing titers in a phase 3 trial of AZD1222 (ChAdOx1 nCoV-19) vaccine²⁷. Our study provides critical insights into protective immunity in a region that has been underrepresented in global SARS-CoV-2 research. Here, we considered Nicaraguan participants with a variety of protective exposure histories (prior infections and various vaccines) and provided mediation analysis results separated by exposure.

Second, we utilized a household-based cohort study with an embedded transmission study design triggered by known virus exposures. Compared to clinical trials or standard observational cohort studies, exposures to infected individuals are expected to be more consistent across participants in household studies and more closely temporally linked to the measurement of antibody titers. For example, in Sun et al., samples were collected several weeks before high incidences of Omicron infections¹⁵. Therefore, the results herein may be more reflective of similar viral exposure challenges across participants than in prior work, and correlates of protection from this study may be useful when considering protection levels at the time of exposure.

Third, we measured titers against the ancestral virus, to which participants were originally exposed earlier in the pandemic and/or by vaccination, as well as against contemporaneous Omicron subvariants BA.1 and BA.2. Relatively few correlates of protection studies have been published with such concurrent analyses of neutralizing antibodies directed against different prior and contemporaneous variants, and none have been household-based cohort with an embedded transmission study triggered by defined index case exposures around the time of titer measurement like this study^{15,22,23,28,29}. Recent COVAIL trial reports similarly identified neutralizing antibodies as correlates of protection against Omicron across mRNA and recombinant protein vaccines^22,23. Our household-based study extends these findings to a community setting where heterogeneous exposures, including multiple infections and vaccines less frequently evaluated globally. One household-based longitudinal cohort study of unvaccinated people in South Africa, without the nested triggering of sample collection that we deployed, measured anti-D614G and anti-BA.1 neutralizing titers months prior to a BA.1 infection wave. In that study, the authors found that BA.1-neutralizing titers mediated protection from infection elicited by prior infections, but also that the difference in neutralizing titer between D614G and BA.1 was itself no longer a correlate of protection, suggesting that homotypic-neutralizing antibody components of ancestral-neutralizing titers are the basis of the nAb correlate of protection¹⁵. Our results showed that protective levels of D614G-neutralizing antibodies are likely higher than protective levels of homotypic-neutralizing antibodies, concordant with the results of Sun and colleagues, in both unvaccinated and vaccinated individuals, and supporting our hypothesis that variant-specific titers would be better correlates of protection. Intriguingly, our GLM-based correlate of protection analysis found that 4-fold increases in ancestral- and homotypic-neutralizing antibodies provided essentially equivalent increases in protection, despite higher average titers across the cohort directed against the ancestral strain. This result may also support the notion that during BA.1 and BA.2 waves, a fraction of cross-reactive ancestral-neutralizing antibodies also had activity against Omicron subvariants. Further, this result points to ancestral-neutralizing antibody titer being a useful correlate of protection even for a viral challenge with later Omicron BA.1 or BA.2 variants.

Our results are also consistent with literature showing that neutralizing antibodies causally mediate protection from infection by SARS-CoV-2 conferred by prior exposures. Here, we further show that neutralizing antibodies mediate protection against infection, and more dramatically, protection against symptomatic infection, conferred by vaccination or prior infection. The cohort studied herein had exposure to global vaccines less frequently studied and most individuals had one or more infections before vaccination.

Finally, we show that serum antibodies, in particular neutralizing antibodies, are clearly informative correlates of protection from symptomatic infection. However, they were correlated with protection from asymptomatic infections only at higher titers, as detected by multiple prospective peri-exposure nucleic acid amplification tests paired with symptom questionnaires. This finding aligns with the pathophysiology of COVID-19 and the known limitations of serum antibodies, as mucosal immunity plays a key role in protection against initial infection³⁰. Thus, while serum antibody measurements are valuable for predicting protection against symptomatic COVID-19, they do not equally correlate with sterilizing immunity against overall SARS-CoV-2 infection.

This study has several limitations. First, there is potential ascertainment bias in our household-based cohort study design, as households with higher susceptibility to viral transmission and disease may have had a greater likelihood of being included due to the occurrence of an index case. Second, while we utilized neutralizing antibody assays to establish protective levels, variations in assay methodologies across different clinical laboratories may influence the exact protective thresholds reported, limiting cross-study comparisons. Third, our analysis to detect correlates of protection for BA.1 infection may be underpowered. Fourth, there may be a difference in the number of unreported SARS-CoV-2 exposures for participants in the earlier BA.1 and later BA.2 waves, which could affect the difference in correlation with protection during these two waves. Fifth, the assumption in our analyses that all household members experienced equivalent exposure to the virus may not fully capture variability in individual exposure risks within households. Sixth, potential confounders such as health-seeking behavior and risk avoidance, comorbidities, host genetic susceptibility, and socioeconomic status (SES) may influence both prior infection or vaccination (the exposure), antibody levels (the mediator), and the risk of infection (the outcome), but were not fully captured in our analysis. Finally, while our causal mediation analysis supports the role of neutralizing antibodies in mediating protection, several key assumptions of the mediation framework, including exchangeability, consistency, and positivity, may not be fully satisfied in this observational context. In particular, the positivity assumption may be violated due to the very low prevalence of individuals without prior SARS-CoV-2 infection in this cohort. This lack of adequate representation of uninfected participants reduces overlap between exposure groups, which can result in unstable or biased mediation effect estimates. Therefore, causal interpretations of these results should be made cautiously.

In conclusion, in this household-based cohort with an embedded transmission study conducted in Managua, Nicaragua, we demonstrated that neutralizing antibody titers, particularly against future strains, are informative correlates of protection from infection and symptomatic SARS-CoV-2 infection. Mediation analyses provide evidence that neutralizing antibodies contribute to protection conferred by vaccination or prior infection, underscoring their role as a mechanistic correlate of immunity. Our findings emphasize the need for continuous surveillance of immune responses to evolving viral variants.