Four statistical frameworks for assessing an immune correlate of protection (surrogate endpoint) from a randomized, controlled, vaccine efficacy trial

Abstract

A central goal of vaccine research is to characterize and validate immune correlates of protection (CoPs). In addition to helping elucidate immunological mechanisms, a CoP can serve as a valid surrogate endpoint for an infectious disease clinical outcome and thus qualifies as a primary endpoint for vaccine authorization or approval without requiring resource-intensive randomized, controlled phase 3 trials. Yet, it is challenging to persuasively validate a CoP, because a prognostic immune marker can fail as a reliable basis for predicting/inferring the level of vaccine efficacy against a clinical outcome, and because the statistical analysis of phase 3 trials only has limited capacity to disentangle association from cause. Moreover, the multitude of statistical methods garnered for CoP evaluation in phase 3 trials renders the comparison, interpretation, and synthesis of CoP results challenging. Toward promoting broader harmonization and standardization of CoP evaluation, this article summarizes four complementary statistical frameworks for evaluating CoPs in a phase 3 trial, focusing on the frameworks' distinct scientific objectives as measured and communicated by distinct causal vaccine efficacy parameters. Advantages and disadvantages of the frameworks are considered, dependent on phase 3 trial context, and perspectives are offered on how the frameworks can be applied and their results synthesized.

Keywords: COVID-19 vaccine; Causal inference; Coronavirus prevention network, COVE vaccine efficacy trial; Mediation; Principal stratification; Surrogate endpoint.

Figure 1.

(A) Hypothetical examples of principal surrogate vaccine efficacy (PSVE) or controlled vaccine efficacy (CVE) curves. Tan shaded regions are densities of the immune marker in vaccine recipients. Interpreted for a naïve-only VE trial cohort, immune markers A, B, C, D are a non-CoP (no value as indicated by a flat curve), a fair CoP, a good CoP (perfect mediator), and an outstanding CoP (perfect mediator and steep S-curve shape), respectively. In naïve-only VE cohorts, PSVE(s1) = PSVE(s1, 0) and CVE(s1) = CVE(s1, 0), where CVE(0) = 0% indicates complete mediation (Non-marker-mediated VE = 0% and PM = 1 as satisfied by markers C and D). (B) Examples of stochastic interventional vaccine efficacy (SVE) curves. Immune markers A, B, C, D are a non-CoP (no value as indicated by a flat curve) and increasingly good CoPs based on increasingly steep curves SVE(d) in d.

Figure 2.

(A) Estimates of principal surrogate vaccine efficacy (PSVE) against COVID-19 by subgroup defined by D29 ID50 titer if vaccinated, with ignorance intervals (dark blue) and 95% estimated uncertainty intervals (light blue) for the sensitivity parameter β set to values in the range [−log(4), 0]. The estimates are plotted for D29 ID50 values ranging from LOD/2 = 1.2 IU50/mL to the 97.5^th percentile value (133 IU50/mL). (B) Estimates of controlled vaccine efficacy (CVE) against COVID-19 under assignment of the whole per-protocol cohort to receive vaccine and to the D29 ID50 titer indicated on the x-axis with 95% confidence intervals (light blue), with the estimates plotted for D29 ID50 values ranging from LOD/2 = 1.2 IU50/mL to the 95^th percentile (80.7 IU50/mL), and the tan shaded region is the estimated density of D29 ID50 titer in the per-protocol correlates cohort of vaccine recipients. (C) Mediation analysis with results reported as point and 95% confidence interval estimates of Marker-mediated VE, Non-marker mediated VE, and the proportion of the overall VE mediated through D29 ID50 titer. (D) Estimates of stochastic interventional vaccine efficacy (SVE) against COVID-19 under hypothetical shifts in D29 geometric mean ID50 titer indicated on the x-axis ranging from a shift of 0 (d=0) (vertical red line) to a one log₁₀ upwards shift (d=1) to a geometric mean of 130 IU50/mL. The horizontal red line indicates the estimated overall VE (against COVID-19 occurring 7 to 126 days post-D29) at the shift of d=0, i.e., the observed marker level. D29, Day 29 visit; ID50, 50% inhibitory dilution; IU50, international units for ID50 titer; LOD, limit of detection; nAb, neutralizing antibody. Panel A is reproduced with minor modifications from Figure S5 (Panel C) in Huang et al. [47]. Panel B is reproduced with minor modifications from Figure S8 (Panel B) in Benkeser et al. [25]. Panel C is reproduced from Table S9 in Gilbert et al. [28]. Panel D is reproduced with minor modifications from Figure S1 (Panel C) in Huang et al. [47].