Application of reverse cumulative distribution curve and scaled logit model in determining optimal immunogenic dose and prediction of protection of EV71 vaccines.

Abstract

Background: This study proposes the reverse cumulative distribution curve (RCDC) for optimal dose selection and a scaled logit model for estimating protection in EV71 vaccine development.

Research design and methods: Data were from a phase 2 trial involving infants and young children randomized to receive either 640 U with or without adjuvant, 320 U with adjuvant, 160 U with adjuvant EV71 vaccines, or placebo. RCDCs were constructed using neutralizing antibody titers 28 days post-vaccination. Robustness of RCDC parameters was assessed via coefficient of variation for the area under the curve (AUC), the relative optimal point, median on the curve, and antibody titer of the point of maximum curvature, with geometric mean titer (GMT) as control. The scaled logit model estimated protection against EV71-associated disease for the selected optimal dose.

Results: The AUC and relative optimal point demonstrated greater robustness than GMT. The 640 U with adjuvant dose had the highest AUC (0.64, 95% CI: 0.61-0.66), sum of coordinates of the relative optimal point (1.40, 95% CI: 1.34-1.43), and the highest estimated protection (93.36%, 95% CI: 79.91-97.86).

Conclusions: AUC and relative optimal point of RCDC are effective for early vaccine dose screening, with protection estimated by the scaled logit model.