Background
The success of modern vaccines has greatly reduced the global burden of infectious diseases, particularly for childhood infections [
1,
2]. However, in sub-Saharan Africa, the childhood burden of malaria remains significant, even with the widespread use of vector control interventions and effective treatment, which have greatly reduced morbidity and mortality. As vaccines are being introduced for more complex diseases, the development of a vaccine for malaria has become a key global health priority. The first strategic goal in the World Health Organization’s Malaria Vaccine Technology Roadmap is the development of malaria vaccines with protective efficacy against clinical malaria of at least 75% over 2 years, for administration to appropriate at-risk groups in malaria-endemic areas, with a booster dose administered no more frequently than annually [
3].
The most advanced vaccine is currently the RTS,S/AS01 vaccine for
Plasmodium falciparum malaria. The phase 3 trial of RTS,S/AS01 was conducted over the period 2009–2014, in two target age groups and for three- and four-dose schedules. In infants aged 6–12 weeks at enrolment who received four doses of the trial vaccine, efficacy was 27.8% (21.7–33.4 95% CI) over a 32-month follow-up period. In 5–17-month-old children who received four doses, efficacy against clinical malaria was 43.9% (39.7–47.8 95% CI) over the same time period [
4]. The RTS,S/AS01 vaccine is unlikely to be pursued as a viable vaccine for infants, due to the low observed efficacy. However, RTS,S/AS01 will now be evaluated through a large-scale pilot implementation program in 5–17-month-old children in three sub-Saharan Africa settings: Ghana, Kenya and Malawi [
5].
Work is ongoing to improve the efficacy of the RTS,S/AS01 vaccine, and recent evidence has indicated that varying the timing and amount of the fourth dose could lead to greater efficacy and improved public health outcomes [
6,
7]. In an RTS,S/AS01 challenge study of healthy adults, with a fractional third dose and fractional booster, efficacy against clinical disease was 86.7% (66.8–94.6 95% CI) at the first challenge (3 weeks after the third dose) and 43% (− 9 to 70 95% CI) at the second challenge (8 months after the first challenge). In the group that received a fractional booster (fourth dose) 8 months after the first challenge, efficacy was 90% (36–98 95% CI) at the second challenge. The immunological reason for this difference is not fully understood, although it may in part be due to improved affinity of the antibodies [
7].
Target product profiles (TPPs) have traditionally been used by industry to guide vaccine and drug development, by setting preferred criteria for product safety, indication, efficacy and cost-effectiveness. However, there is increasing focus on using TPPs as more adaptable, broader tools that capture the full public health value of a drug or vaccine, to help a wider range of stakeholders, such as policy-makers, design and evaluate vaccine formulations [
8]. For diseases with complex epidemiological features, such as malaria, mathematical modelling can be particularly useful for informing the public health impact components of TPPs [
9].
In this study, we used mathematical models of malaria transmission and vaccine efficacy to predict the impact of childhood vaccination with a modified RTS,S/AS01 vaccine and to inform TPPs for second-generation vaccines, focussing on the elements of initial efficacy, duration of protection, dosing schedules and coverage [
10‐
13]. We determined the relative importance of initial vaccine efficacy versus duration of protection in a range of
P. falciparum prevalence settings, in terms of clinical cases averted in children younger than 5 years, for the first decade following vaccine introduction. We also considered how the timing and efficacy of the fourth dose are likely to change the public health impact. Finally, we explored options for realistic vaccine efficacy profiles for a modified RTS,S/AS01 vaccine, in line with the results of a recent study by Regules et al. [
6], and used these profiles to estimate the impact of an enhanced vaccine in different transmission settings.
Discussion
In our study we used a dynamic modelling approach to compare the relative contributions of vaccine characteristics, to inform a TPP for a second-generation malaria vaccine for young children. We first examined the characteristics of initial efficacy and duration in the context of a three-dose schedule before analysing the characteristics specific to the fourth vaccine dose. Our results demonstrated that, for the three-dose vaccine schedule, initial efficacy was associated with a higher reduction in childhood clinical cases compared to vaccine duration. This effect was more pronounced in high malaria transmission settings due to the concentration of clinical cases in younger age groups, and this finding held true when a fourth vaccine dose was incorporated (where the fourth dose had the same efficacy and timing across all scenarios). However, we found that there was a larger age-shifting effect in higher transmission settings, where the initial efficacy was higher, due to the delayed acquisition of clinical immunity [
17].
Examining the characteristics of the fourth dose, we found that the largest benefit in terms of clinical cases averted always occurred when the efficacy at the fourth dose was highest, and that a longer timing between doses three and four was also advantageous. In the third part of the analysis, we created three scenarios where all of the vaccine parameters — initial efficacy, duration and fourth dose titre and timing — were varied, to replicate the RTS,S/AS01 phase 3 profile; a profile approximately corresponding to the fractional dose study (referred to as the modified profile); and an intermediate profile. This analysis showed that a modified malaria vaccine would likely outperform the current RTS,S/AS01, but that increasing the timing between doses three and four could further reduce the number of clinical cases in older children.
The World Health Organization has specified that a malaria vaccine for children younger than 5 years should reduce incidence of all clinical malaria episodes by at least 75% for at least 1 year and preferably at least 2 years [
3]. Importantly, the Preferred Product Characteristics state that initial efficacy and duration will be jointly considered, and the duration of protection is assessed as being as important as the short-term efficacy for the target vaccine group, in medium to high transmission settings [
18]. Our findings indicate that, for imperfect vaccines with suboptimal efficacy, it may be advantageous to prioritise initial efficacy over duration.
The results of our study should be interpreted in the context of the selected follow-up time window, age cohorts and outcome measure. We focussed on children younger than 5 years and a 10-year horizon following vaccine introduction as our primary outcome measure, given the circumstances in which a childhood vaccine is likely to be assessed in field studies, as well as the many other changes to malaria interventions likely to occur over longer time periods. We found that the low initial efficacy and long duration profile resulted in more clinical cases averted when measuring cases in all age groups, and over a longer time window; however, this is likely due to the functional form of this efficacy profile allowing the vaccine benefit to extend for far longer than expected in practice. We also found that measuring the severe cases averted across all ages meant little difference in impact of the three efficacy profiles considered, further justifying the focus on childhood impact. Finally, our main outcome measure was clinical cases averted, and we did not account for the implementation costs associated with different schedules, which may change how vaccine efficacy and duration are prioritised.
Our study had several limitations. First, we fixed the age at first vaccine delivery at 5 months and did not incorporate any reduction in susceptibility to infection until after the third vaccine dose. It is possible that some protection is afforded after the first vaccine dose, which would further increase the vaccine impact. Additional data on the protection between doses would be needed to capture this in the modelling framework. Second, our proxy for total efficacy, i.e. total area under the time-efficacy curve over a 10-year period, meant that some of the area under the efficacy curve was outside the measured time window. However, our study focussed on impact in the childhood age group, and so we do not expect this assumption to greatly influence our results.
Our method assumes that there is a non-linear but positively increasing relationship between antibody titre and vaccine efficacy. This is biologically plausible based on findings from the RTS,S/AS01 vaccine. However, in the phase 2a challenge study there was not a great change in antibody levels between the protected and non-protected groups, suggesting that factors other than antibody titre may influence efficacy [
6]. By varying the antibody titre at the fourth dose, we were able to capture the observed challenge study efficacy, without explicitly incorporating other factors into our antibody model, and these efficacy curves in turn informed the estimates of population-level vaccine impact. However, the efficacy of a second-generation vaccine may need to be improved by enhancing antibody quality or avidity, rather than by solely increasing antibody titre (given that higher titres may not be achievable in African children). Fitting of dose-response relationships to data from challenge studies for other malaria vaccines could be undertaken to incorporate other factors such as antibody quality into this modelling framework, to consider the potential public health impact of candidate vaccines in phase 2 studies.
Our analysis was motivated by the findings of the fractional third dose and fractional booster dose study by Regules et al. [
6], which provides an early indication that changing dosage and timing could improve vaccine efficacy. However, the fractional dose study was conducted in a relatively small sample of 30 healthy malaria-naïve adults, meaning there are limitations in comparing these results with those from the phase 3 trial. In addition, it is not clear from this single study whether the increased efficacy is due to the fractional third dose or to the delay in the third dose [
7]. A phase 2 trial of RTS,S/AS01E compared 0, 1, 2 month and 0, 2, 7 month dosing schedules within the Expanded Programme on Immunization. The trial found that over the 19 month study period, efficacy was higher in the 0, 1, 2 month group, although when comparing malaria episodes in the 12-month period following the third dose, efficacy across the two schedules was similar (noting that this study was in infants only) [
19]. Further characterisation of the dose-response relationship between antibody titre and efficacy within this and future challenge studies could help elucidate this mechanism and guide TPP development. Furthermore, in creating the ‘intermediate’ vaccine efficacy profile, with parameters approximately at midpoints between those selected for the phase 3 and modified efficacy profiles, we implicitly assumed a trade-off between the timing of and antibody titre at the fourth dose, where a later dose timing resulted in a lower antibody titre (and therefore efficacy) following the fourth dose. A more complex relationship between these factors is possible and could be quantified given further data on this relationship.
Despite the success in developing the RTS,S/AS01 vaccine, there have been challenges, many of which will impact the progress of a second-generation vaccine. A key issue is that we still do not have a full understanding of the immunological mechanisms by which the RTS,S/AS01 confers protection against malaria disease, and there is no widely accepted immune correlate of protection. While there is evidence that high anti-CSP antibodies are associated with protection, antibody titre is not an established protective correlate. Further, it is likely that a second-generation vaccine will need to be assessed relative to the first-generation vaccine, which presents feasibility issues in terms of community acceptance, trial design and trial sample size. The measures by which the success of a new vaccine is assessed will also need to be carefully considered [
20,
21].
Conclusions
In vaccine development and evaluation, considerable focus is typically given to the durability, or half-life, of a vaccine. However, our findings show that these vaccine properties need to be considered in the context of the target population and immunological profile. In our study, we found that in the first decade of vaccine implementation the initial efficacy was more important than duration, based on our current understanding of the RTS,S/AS01 vaccine and in terms of childhood clinical cases averted. This study also emphasises the importance of considering the age distribution of incidence in a range of disease prevalence settings when predicting the impact of an age-targeted vaccine. Finally, our analysis demonstrates that the timing of the fourth RTS,S/AS01 dose can change the overall health benefit, which should be considered with the logistics of implementing a childhood malaria vaccine within the current Expanded Programme on Immunization framework. The findings from this analysis could provide insight for vaccine developers and policy-makers into how distinct properties of a malaria vaccine may translate to public health outcomes and how the importance of these characteristics changes across different malaria prevalence settings.