Dengvaxia Efficacy Dependency on Serostatus: A Closer Look at More Recent Data

To the Editor—We read with concern the recent article by Yang and colleagues, “Dependency of Vaccine Efficacy on Pre-Exposure and Age: A Closer Look at a Tetravalent Dengue Vaccine” [1]. We agree that Dengvaxia was efficacious among children who were dengue seropositive at baseline, but it is important to mention that it is not known how long vaccinated seropositive individuals will be protected. More studies are necessary as this vaccine could be giving only a temporary boost to their immunity. The results from Yang et al focus, inappropriately, on data from the active phase of the trials only, not considering the information regarding the long-term safety of the CYD-TDV vaccine, but draw wide conclusions, such as “vaccine efficacies among 6–8 years old are comparable to those among 9–11 years old, regardless of baseline immunity status” and that “children in this age group in dengue-endemic countries could be considered for vaccine coverage.” The authors have performed a modified intent-to-treat analysis by excluding some individuals, showing only the estimates obtained by statistical analysis, but not the raw numbers used in their analysis. Vaccine trial data for the follow-up years are available since 2016 [2], also by individual serostatus [3], so the results of Yang et al do not reflect the present knowledge about the performance of Dengvaxia. An analysis of recent publicly available data on age and serostatus confirmed statistically a vaccine-induced risk in seronegative individuals (see Figure 1). Using a more recent data set by age and serostatus from the combined CYD14, CYD15, and CYD57 trials, as reported in Table 1 in Martínez-Vega et al [3], the overall vaccine efficacy (VE) for hospitalized virologically confirmed dengue cases was estimated via the Bayesian approach [4–6] to obtain a probability p(k|Iv,Ic) for VE k with infected individuals Iv in the vaccine group and Ic in the placebo group. The Bayesian estimate for seropositive individuals 2–8 years old is k = 60% (95% confidence interval [CI], 4.8%–8.4%). Data for seropositive individuals aged >9 years were not available. For seronegative individuals 2–8 years old, the estimate is negative, k = −76% (95% CI, −421% to 29%), whereas the estimate for seronegative 9- to 16-year-olds is slightly positive, but not statistically significant for positive or negative VE, k = 8% (95% CI, −235% to 71%). For all seronegative individuals combined, 2–16 years old, VE is again negative, k = −40% (95% CI, −215% to 32%), hence with a relative risk of hospitalization larger 1 for vaccinated seronegatives at any age enrolled in the trials, as compared to the individuals in the control group. The comparison between VE distributions by serostatus, for seronegative and seropositive individuals sharing the same age, shows a significant difference among those individuals (Figure 1A), with little overlap observed. However, regardless of age differences, a large overlap between the VE distributions of seronegatives of those aged 2–8 years and 9–16 years (Figure 1B) is observed, suggesting that serostatus is determining the efficacy of this vaccine and not age. As for the cumulative distribution for all seronegative individuals, aged 2–16 years, enrolled in the vaccine trials (Figure 1C), we observe 82% probability of negative VE (ie relative risk [RR] larger 1) but only 18% probability of positive VE (ie RR smaller 1), indicating that the individual immunostatus prior to vaccination needs to be considered. Bayesian estimate of the overall vaccine efficacy (VE) k for hospitalized confirmed dengue cases, by serostatus and age. Data from Table 1, as reported in [3], are used in the VE estimate distribution by serostatus for individuals aged 2–8 years (A). The red curve shows the estimates for the seronegative individuals, with the median shown in black. The purple curve shows the estimates for seropositive individuals. Little overlap on the distributions is observed, indicating significant differences in respect to individual serostatus. B, VE estimate distribution for seronegatives in different age groups. The red curve shows the estimates for the seronegative individuals aged 2–8 years, whereas the blue curve shows the estimates for the seropositive individuals aged 9–16 years. Here, large overlap between the distributions is observed, an indication that serostatus is more important to determine the efficacy of this vaccine than age. C, Cumulative VE distribution for all seronegative individuals, aged 2–16 years, enrolled in the CYD15/CYD15/CYD57 vaccine trials. More than 80% probability for negative VE is encountered for seronegative individuals. Dengvaxia is now licensed for use in several countries, and we urge that all vaccination sites choose the most appropriated screening test to identify seropositive individuals, independent of age, minimizing possible vaccine disease enhancement among seronegative individuals [7, 8]. Financial support. This work was supported by Strategic Projects UID/MAT/00297/2013 (Centro de Matemática e Aplicaçẽs, Universidade Nova de Lisboa) and ID/MAT/04561/ 2013 (Centro de Matemática, Aplicaçẽs Fundamentais e Investigaccão Operacional, Universidade de Lisboa). Potential conflicts of interest. Both authors: No reported conflicts of interest. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.