Identification of RTS,S/AS01 vaccine-induced humoral biomarkers predictive of protection against controlled human malaria infection

Abstract

BACKGROUNDThe mechanism(s) responsible for the efficacy of WHO-recommended malaria vaccine RTS,S/AS01 are not completely understood. We previously identified RTS,S vaccine-induced Plasmodium falciparum circumsporozoite protein-specific (PfCSP-specific) antibody measures associated with protection from controlled human malaria infection (CHMI). Here, we tested the protection-predicting capability of these measures in independent CHMI studies.METHODSVaccine-induced total serum antibody (immunoglobulins, Igs) and subclass antibody (IgG1 and IgG3) responses were measured by biolayer interferometry and the binding antibody multiplex assay, respectively. Immune responses were compared between protected and nonprotected vaccinees using univariate and multivariate logistic regression.RESULTSBlinded prediction analysis showed that 5 antibody binding measures, including magnitude-avidity composite of serum Ig specific for PfCSP, major NANP repeats and N-terminal junction, and PfCSP- and NANP-specific IgG1 subclass magnitude, had good prediction accuracy (area under the receiver operating characteristic curves [ROC AUC] > 0.7) in at least 1 trial. Furthermore, univariate analysis showed a significant association between these antibody measures and protection (odds ratios 2.6-3.1). Multivariate modeling of combined data from 3 RTS,S CHMI trials identified the combination of IgG1 NANP binding magnitude plus serum NANP and N-junction Ig binding magnitude-avidity composite as the best predictor of protection (95% confidence interval for ROC AUC 0.693-0.834).CONCLUSIONThese results reinforce our previous findings and provide a tool for predicting protection in future trials.TRIAL REGISTRATIONClinicalTrials.gov NCT03162614, NCT03824236, NCT01366534, and NCT01857869.FUNDINGThis study was supported by Bill & Melinda Gates Foundation's Global Health-Discovery Collaboratory grants (INV-008612 and INV-043419) to GDT.

Keywords: Immunology; Malaria; Vaccines.

Figure 1. Schemas and study schedules.

(A) Schematic of different domains in PfCSP. The major repeats of the NANP motif are shaded in gray and the interspersed minor repeats of the NVDP motif are shaded in blue. The location of the N-terminal junction region of the CSP-containing NPDP motif (shaded in red), which is referred to here as N Interface, is indicated. The portion of CSP included in RTS,S vaccine is indicated. Schedule of vaccination, CHMI, and serum draws tested for humoral immunity are shown for MAL092 and MAL102 (B), MAL068 (C), and MAL071 (D). *The MAL092 Adu1Fx group was not vaccinated on day 28.

Figure 2. Predictive value of previously identified antibody biomarkers in an independent study population.

ROC curves quantifying the ability of biomarkers individually and in combination to predict protection based on day of CHMI data are shown. Predictive models were trained using day of CHMI data from MAL068 and MAL071 combined and used to predict protection in MAL092 (A) and MAL102 (B), trained using day of CHMI data from MAL092 and MAL071 combined and used to predict protection in MAL068 (C), or trained using day of CHMI data from MAL092 and MAL068 combined and used to predict protection in MAL071 (D).

Figure 3. Serum Ig, IgG1 CSP, and NANP6 measurements associate with protection status in MAL092.

Odds ratio estimates (circles) and 95% CIs (error bars) obtained from regimen-adjusted logistic regression models fit independently to each immune measurement (Protection ~ Regimen + immune measurement) based on MAL092 DoC data are shown. Asterisks represent statistically significant associations based on FDR-adjusted (Benjamini-Hochberg) P values: *P < 0.05, **P < 0.01.

Figure 4. Serum Ig, IgG1 CSP, and NANP6 measurements associate with protection status in MAL092.

Antibody measurements at DoC in the MAL092 study that have statistically significant associations with protection status based on regimen-adjusted logistic regression models fit independently to each immune measurement are shown. CSP (A), NANP6 (B), and N Interface (C) AUC_diss, and IgG1 CSP (D) and NANP6 (E) 334-hIgG1 equivalent concentrations are compared for protected and nonprotected vaccinees (labeled as P and NP, respectively). Raw and FDR-corrected (Benjamini-Hochberg) P values shown from regimen-adjusted logistic regression models fit independently to each immune measurement. n = 94 (53 protected and 41 nonprotected). The lower and upper hinges of the box-and-whisker plots correspond to the 25th and 75th percentiles, with a line at the median. The lower and upper whisker extends from the box hinges to the smallest and largest values, respectively, which are within 1.5 × IQR of the hinge (where the IQR is equal to the distance between the 25th and 75th percentiles).

Figure 5. Serum NANP6 Ig, serum N Interface Ig, and IgG1 NANP6 measurements combined associate with protection status.

Mean ROC curve (solid line) and 95% CI (shaded area) are plotted for best predictive model identified by 10-fold cross-validated logistic regression with LASSO penalty based on DoC data from MAL068, MAL071, and MAL092 combined.

Figure 6. One-year durability of RTS,S/AS01-induced antibody responses.

Fold change in antibody responses from MAL092 day 226 (30 days after the final immunization) to MAL102 day 1 (day of boost or 1 year after the final MAL092 immunization) versus MAL102 day 1 antibody responses by MAL092 protection status. PfCSP-specific (A), NANP6-specific (B), and N Interface–specific (C) AUC_diss, and IgG1 CSP (D), IgG1 NANP6 (E), and IgG3 CSP (F) 334-hIgG1/hIgG3 equivalent concentrations. For MAL102 participants who were protected (P) or nonprotected (NP) from MAL092 CHMI, individual data points represent the median and error bars represent 25th to 75th percentiles. Horizontal dotted line represents no change in antibody response from MAL-092 day 226 to MAL-102 day 1. Asterisks along the horizontal and vertical axes indicate significant P values from Mann-Whitney U tests: *P < 0.05, **P < 0.01. NS, not significant.