Abstract / Summary
RTS,S/AS01 (RTS,S) is one of two WHO-recommended malaria vaccines for children in moderate-to-high transmission settings. Yet, no validated immune correlate of protection (CoP) has been established for this vaccine. We selected 16 biomarkers to further assess as candidate CoPs, based largely on their ability to well-predict post-challenge outcomes in a previous integrative correlates analysis of two phase 2a controlled human malaria infection (CHMI) studies, Malaria-068 and Malaria-071. We evaluated the predictive power of these 16 biomarkers in two additional phase 2a CHMI studies, Malaria-092 and Malaria-102, to confirm these candidate CoPs in new malaria vaccine trials. In MAL092, malaria-naive volunteers were challenged three months after a series of two or three doses of RTS,S/AS01, and in MAL102, a subset were boosted nine months later and re-challenged after three weeks. The 16 biomarkers included Plasmodium falciparum circumsporozoite protein (CSP)-specific IgG antibodies targeting the CSP NANP-repeat region (CSP IgG ELISA), NANP repeat-specific antibody binding to human Fc gamma receptors (FcgRs), CSP-specific antibody Fc effector functions, total NANP6-specific serum Ig measured by biolayer interferometry (BLI), and NANP6-specific IgG1 antibodies measured by binding antibody multiplex assay. Immunogenicity analyses confirmed that the vaccine regimens elicited robust immune responses, including multiple biomarkers not previously reported on, in both trials. In MAL092, participants clustered into two distinct immunogenicity profiles that overlapped with malaria protection status. In univariate logistic regression analyses, most biomarkers were significantly associated with protection when measured on the day of challenge in both trials. Notably, the same three biomarkers ranked as top correlates in both trials (FCGR2AH NANP6, FCGR2AR NANP6, and FCGR2B NANP6; all adjusted p-values <0.001), with similar effect sizes across studies. Despite these strong associations, no biomarker demonstrated significantly better individual-level predictive performance than CSP IgG ELISA. In MAL092, CSP IgG ELISA achieved an area under the curve (AUC) [95% confidence interval (CI)] of 0.74 (0.63, 0.84), compared with AUCs of 0.76 - 0.78 for the top five biomarkers, all of which reflected NANP6-specific Fc{gamma}R binding (all adjusted p-values > 0.37 versus CSP IgG ELISA). Predictive models trained on data from two prior CHMI studies and applied to MAL092 and MAL102 identified a model based on FCGR2B NANP6 (validation AUC = 0.77 in MAL092; 0.80 in MAL102) as the best-performing univariate model for both studies. The best-performing multivariate predictive model for MAL092 was based on BLI NANP6 Dissociation AUC and ADCP NANP6 (validation AUC = 0.76), and in MAL102, it was based on CSP IgG ELISA, BLI NANP6 Dissociation AUC, and ADCP NANP6 (validation AUC = 0.82). The next approximately 10 highest-ranked models, however, had comparable performance with overlapping 95% CIs, and none were clearly superior to CSP IgG ELISA (validation AUCs 0.74 and 0.71 in MAL092 and MAL102, respectively). Collectively, these findings suggest that the majority of the previously identified candidate CoPs also hold up well in two additional malaria vaccine CHMI trials and underscore the continued utility of the CSP IgG ELISA readout as a robust marker of protection. Further studies are needed to determine whether these candidate CoPs hold in malaria-endemic field settings. Identifying validated CoPs that predict malaria vaccine efficacy would promote the development and approval of more effective and durable next-generation vaccines. These advances could also inform vaccine strategies for other infectious diseases.