Improving precision of vaccine efficacy evaluation using immune correlate data in time-to-event models

Figures

• 

  Fig. 1. Zoster vaccine is generally less immunogenic in older subjects than in younger subjects. Distribution of log2 fold rise in VZV antibody titers (gpELISA) at baseline and week 6 (i.e., 6 weeks post vaccination) in HZ cases (red circles) occurring between week 6 and end of follow-up (~3 years), and non-cases including all participants who did not have HZ during the follow-up25. Boxplots show first quartile (lower edge of the box), median (horizontal line in the box), and third quartile (upper edge of the box), with participants stratified according to treatment assignment and HZ disease status; the whiskers indicate 1.5 times the interquartile range from the box. Red and gray circles show individual values of log2 fold rise in VZV antibody titers of HZ cases and non-cases, respectively.

• 

  Fig. 2. Estimated relationships between log2 fold rise in VZV antibody titers and hazard ratio (with median immunogenicity observation as the reference) of HZ, ϱ(T). Fold rise is an inverse correlate of risk (i.e., is significantly associated with hazard of HZ) in both younger and older subjects. Log2 fold rise is the only predictor in the selected model (Model a, Supplementary Table 1), risk curves for younger and older groups are therefore identical. The hazard ratio (y-axis) represents the hazard relative to the median of all immunogenicity data (shown with black vertical line) for which it equals to 1. The bottom plots show individual values of log2 fold rise by age group and vaccination status (gray points) with respective empirical probability density functions; gray vertical lines represent median, first quartile, and third quartile. For all placebo recipients, log2 fold rise is assigned to be 0 (see Methods); the density function shown in the plot for placebo recipients is an artifact of the smoothing function used by the smoothing kernel density estimator.

• 

  Fig. 3. Immunogenicity of CYD-TDV, scaled and centered average log2 PRNT50 titers. Average titer was determined for each subject as the arithmetic average of respective four centered and scaled serotype-specific log2 PRNT50 titers. For each serotype, first, centering is performed by subtracting the mean log2 titer (across all subjects) from each log2 titer observation (per subject). Next, scaling is performed by dividing each (centered) log2 titer observation (per subject) by the standard deviation of log2 titers (across all subjects). Boxplots show first quartile (lower edge of the box), median (horizontal line in the box), and third quartile (upper edge of the box), with participants stratified according to treatment assignment and serotype-specific VCD endpoint; the whiskers indicate 1.5 times the interquartile range from the box. Red and gray circles show individual average antibody titers of serotype-specific VCD cases and non-cases, respectively.

• 

  Fig. 4. Estimated relationships between hazard ratio, ϱ(T), and average log2 PRNT50 titer (with median immunogenicity observation as the reference) of serotype-specific VCD. Risk curves differ significantly between seropositive and seronegative subgroups for DENV-Any, DENV1, and DENV2. For DENV-Any and DENV2 the Cox PH model with quadratic term for average titer (non-monotonic risk curves) provided the best fit to data; the increase in risk with average titer is not significant, because confidence intervals include a constant curve over the titer range for non-monotonicity. The hazard ratio (y-axis) represents the hazard relative to that for a population with titer at the median of all immunogenicity data (observed in a seropositive subject and shown with black vertical line) for which (by definition) this ratio equals 1. The bottom plots show (gray points) individual participants’ values of average titer (averaged across serotypes after mean-centering and scaling by standard deviation) by serostatus (column) and vaccination status (row) with respective empirical probability density functions; gray vertical lines represent median, first quartile, and third quartile. For seronegative…