Response to Comment on ‘Evaluating potential associations between prior maternal yellow fever vaccination and protection against infant adverse outcomes following Zika virus antenatal exposure’

We thank Drs Lin and Gang for their thoughtful comments and for highlighting the relevance of exploring how prior Yellow fever (YF) vaccination may modify susceptibility to Zika virus (ZIKV) infection and adverse infant outcomes [1]. Our study was not restricted to microcephaly. We examined whether yellow fever neutralizing antibody (YF NAb) titers were associated with protection against maternal ZIKV infection and, among ZIKV-exposed pregnancies, whether higher YF NAb titers were associated with a lower risk of early or late adverse infant outcomes.

Regarding sample size, Tables 1 and 2 present descriptive clinical characteristics for all eligible mother–child dyads with available clinical data (n = 146), derived from the parent cohort [2] (n = 157) after exclusion of pregnancy losses. YF neutralization assays were performed only among mothers with available serum samples (n = 129), as stated in the “Methods” section [1]. Of these, nine mothers who were ZIKV-positive by plaque reduction neutralization test (PRNT) but had indeterminate timing of infection during pregnancy were excluded a priori from inferential analyses, resulting in a final analytical sample of 120 participants. We acknowledge that this distinction between descriptive and analytic denominators was insufficiently explicit in the tables and apologize for this oversight.

We also acknowledge a reporting error in the “Results” section, in which mean and standard deviation values comparing YF Nab titers between ZIKV-exposed and unexposed mothers were incorrectly transcribed during manuscript preparation. The underlying analyses were conducted using the correct data. Using the original analytical dataset, mean YF NAb titers were higher among ZIKV-unexposed mothers (n = 53; mean, 461.4, SD, 605.4; median, 242.0, IQR, 112.0–482.0) than among ZIKV-exposed mothers (n = 67; mean, 225.4, SD, 276.7; median, 136.0, IQR, 79.0–288.0). Both Welch’s t-test (P = .011) and the Mann–Whitney U-test (P = .029) indicated consistent differences between groups. We regret this reporting error and emphasize that it does not affect the statistical inference or interpretation of study findings.

With respect to modelling strategy for long-term outcomes, we agree that the limited number of events constrains model complexity. As stated in the “Methods” and “Discussion” sections [1], our analyses were exploratory and hypothesis-generating rather than causal, with automated selection used for parsimonious adjustment, consistent with methodological recommendations [3-5]. In response to this comment, we conducted an additional sensitivity analysis using a prespecified multivariable logistic regression model that retained YF NAb titers a priori, modeled on a log10 scale, and including family income as a continuous socioeconomic indicator, without automated variable selection. Consistent with the main analysis, log-transformed YF NAb titers were not independently associated with long-term neurodevelopmental delay (OR, 0.99; 95%CI, 0.38–2.61), nor was family income (OR, 0.99; 95%CI, 0.99–1.00). Although estimates were imprecise due to a limited number of events, the direction and interpretation of findings were unchanged.

Finally, we agree that potential confounding by other flavivirus exposures, particularly dengue virus infection, warrants careful consideration. The primary aim of our study was to assess prior immunity to YF using a highly specific neutralization assay (μFRNT-HRP), the gold standard for confirming YF virus–specific NAbs [6, 7]. Evaluation of interactions with other flaviviruses or distinction between vaccine and infection-induced immunity was beyond the scope of this investigation. Our findings are nevertheless consistent with population-level data from Brazil showing lower ZIKV attack rates in areas where YF vaccination was part of the routine immunization schedule during the Zika epidemic, supporting biological plausibility [8]. We emphasize that this was a pilot, hypothesis-generating study intended to prompt further research, while fully supporting current YF vaccination recommendations.