Given the enormous multiplicity of combinations of interventions that might be considered when targeting the whole population with combination strategies, and given the expense of mass deployment, it is critical to carry out in silico investigations of the potential of novel deployment options in order to avoid wasteful field trials of fruitless combinations .
To date little has been done to estimate the impact of adding a similar AIV to existing mass treatment strategies. By combining mass vaccination to treatment, the infectious reservoir is reduced by clearing infections in a large proportion of the population via MDA, and adding an intervention with ability to prevent infections for longer than the duration of the transmission season (even with waning protective immune responses restored with yearly immunizations) resulted in higher maximum prevalence reduction, increased chance of elimination, and importantly, delayed resurgence compared to interventions without the vaccine. Our simulations suggest that combining MDA with an AIV has a synergistic effect in regards to chances to interrupt transmission, and synergism increased with greater levels of initial prevalence and lower MDA coverage or case-management. Synergism can be the result of deploying two interventions which increases overall coverage of the population covered by at least one intervention, but as synergism was also predicted when the vaccine and the drug were given to the same people given coverage, although in a lesser extent, it is also likely a consequence of the interventions targeting different parasite life stages.
It will be important to consider whether the approaches modelled here are operationally feasible and what level of health system and program strength is required to coordinate vaccination and drug campaigns. For this reason, we investigated a range of coverages and our main results were presented for coverages of 60%, assuming any higher was too optimistic. In those conditions, with initial PfPR2–10 = 4%, we found that for the simulations where transmission is not interrupted, the rate of resurgence is halved compared to MDA alone, extending the resurgence half-life up to two and a half years. Alternatively, we investigated a higher MDA coverage compared to vaccination and we found that adding mass vaccination even at a lower coverage (40%) remained beneficial in delaying resurgence and reducing MDA rounds. We also found limited differences in impact between the combined strategies with vaccination before or during the transmission season, arguing for a preference of during the season with 3 visits as opposed to more visits.
In addition, we estimated that vaccines of longer duration of protection can further delay resurgence, arguing that research into the feasibility of increasing longevity of protection of a vaccine has importance for the elimination strategies investigated here.
The addition of mass vaccination to MDA, compared with MDA alone, has the potential to reduce the number of MDA rounds needed, as well as the number of years that interventions may be needed, to achieve similar public health benefits of prevalence reduction and chances to interrupt transmission, and thus reduce the length of deployment of interventions from 3 to 2 years. Reducing the length of overall intervention deployment has the potential to reduce the number of mass intervention visits, and thus costs of visits. Moreover, although an economic analysis was outside the scope of this study, a shorter period of deployment could be beneficial to national malaria control and elimination strategies facing budgetary constraints. Although this work did not explicitly explore mobility, we did model importations of infections and highly mobile populations will limit operational coverage of the vaccines and MDA. Limited duration mass interventions are not intended to tackle importation. As with MDA, importation is likely to have little effect on the estimated prevalence reduction unless initial prevalence is very low, and in these settings where interruption of transmission is achieved a high level of case management will be important to maintain zero prevalence . In line with this, we found importation of infections had little effect on prevalence reduction achieved but that time to resurgence decreased as importation increased.
Furthermore, interruption of transmission with mass interventions is likely only possible for scenarios of low initial prevalence (low prevalence as a result of vector control or other interventions), with a high level of case management and an increase in other intervention tools, which was recently highlighted by the WHO Malaria Policy Advisory Committee meeting with regards to the use of MDA . Our simulations also indicated that, chances to interrupt transmission increase with higher case management. This implies that such mass interventions should not be investigated independently of case management, and that together with vector control or reactive tools, will play a decisive role in the success of the interventions to reach elimination and avoid resurgence.
This work was confined to a limited number of model-based scenarios of mass interventions and as such there are several limitations. Firstly, several aspects not captured might enhance or diminish the impact of mass vaccination combined with MDA. On a biological level, vaccination with an AIV with drugs has not been sufficiently investigated, and it is unclear how vaccine efficacy, immune dynamics in response to the vaccine, the length of protection, or the rate of vaccine failure would be influenced. Such effects are unknown, however a lower rate of vaccine failure and longer duration of protection would delay resurgence and extend the pre-elimination period, and conversely higher failure rates would decrease benefits. Additionally, we did not address any potential evolutionary effects such as vaccine insensitivity, nor antigenic sin. However, we note deploying with treatment may protect against evolution of vaccine insensitivity.
Secondly, the underlying vaccine assumptions in the models, including the effect of a forth vaccine dose, may influence results. Vaccine properties were based on analysis of the results of RTS,S Phase 3 clinical trials , which targeted infants aged 6 to 12 weeks and young children aged 5 to 17 months. In the context of mass vaccination, as the efficacy and protection might be different for different ages and previous exposure, extension of the protection profile and vaccine parameterization from children to the entire population is a necessary simplification based on currently available data. A full sensitivity analysis around vaccine efficacy profiles has not been performed in this study, but simulations where initial efficacy against infection for children and adults was reduced to 50% suggest that this would considerably decrease the prevalence and case management settings where the combination strategies would reach very low transmission levels or interruption of transmission.
Thirdly, we did not model potential increases in vector control or case management levels over the time of intervention, either due to vector control campaigns or active efforts to improve access to treatment or due to a better efficiency in the health system as fewer cases need to be treated. Both would further reduce prevalence levels following mass intervention deployment, and in some settings, would likely enhance the effect of mass vaccination, especially during the resurgence period. Conversely, we did not model deceasing levels of vector control which may decrease the benefits of the mass intervention strategies or increase rates of resurgence.
We chose a 3-month seasonal malaria transmission setting of sub-Sahel African countries approximating the pattern in Senegal where SMC is ongoing, and investigated the impact on average transmission, not accounting for heterogeneities in transmission. Further studies could be undertaken to tailor the analysis specifically to a country or region, with corresponding transmission pattern, heterogeneity, and other underlying settings, such as case management or adapted timing and duration of deployment. The impact of interventions may be sensitive to timing and the seasonal pattern chosen for malaria transmission, and that the heterogeneity of transmission across a given region will imply a need for focal planning.
Previous modeling has shown the impact of MDA is very sensitive to population size, being best suited for small populations and with total impact sensitive to the proportion of the population receiving no treatment from any of the deployed rounds . In our current work, individual rounds of MDA were given to a random proportion of the population for a given coverage, mass vaccination was deployed either independently to MDA or in some settings simultaneously to MDA, and the 3 vaccine doses independently from the fourth dose. As expected, scenarios with total correlation between those covered by MDA and mass vaccination predicted a slightly reduced impact, but the key determinants probably remains in the random coverage between MDA rounds similar to results found for multiple rounds of MDA , and total correlation between rounds has not been investigated here. In more realistic settings the proportion of the population covered would neither be completely independent nor would it be exactly the same between deployments, and thus our assumptions might be slightly optimistic so that the nominal coverages we simulate reflect higher values in the field.
Given that MDA and SMC are currently implemented in several settings [19, 40], and that the combination of RTS,S and SMC is being tested in field trials , we are in a position to consider adapting these combination of interventions to different ages. Technical challenges however remain. It is envisioned that RTS,S may be delivered in much the same way as meningococcal A conjugate vaccine is being deployed in sub-Saharan Africa in 1–29 year old, initially in intense mass vaccination campaigns, and then followed by introduction into existing Extended Immunization Programmes in the individual countries . Any vaccines used in malaria elimination campaigns will need to be WHO pre-qualified ensuring that characteristics are aligned with cold-chain requirements available in the countries where it is deployed . For an AIV to be used as a mass intervention in Africa or Asia, several important unknowns must be addressed with respect to immunogenicity in adults and immunogenicity in populations outside Africa. Measurement of immune responses and protection in adults have been tested in the Mekong, where RTS,S was administered alone or in combination to drugs  as are many other CHMI (controlled human malaria infection) studies of RTS,S, R21, and PfSPZ. These studies should provide additional evidence that could be used to model their potential role in mass vaccination scenarios. Other challenges include safety and efficacy testing when vaccines are administered with drugs and regulatory hurdles for use of any of the AIVs for children and adults, which may be more difficult with new antigens. From an operational perspective, the challenges may be significant, ensuring sufficient coverage of the first three or required initial number of doses of the vaccine. And lastly, to make use of the potential benefit in delaying resurgence, settings must be prepared with additional interventions during the low prevalence period following 2 or 3 years of mass intervention, and as with MDA, it will be important to include surveillance and response strategies combined with strong health systems to address importation of infection and preserve this delay in resurgence.
Given combining mass vaccination with treatment has potential to delay resurgence for longer than treatment alone, there may be a role in malaria outbreaks or in fragile contexts where health systems are weakened by other disease outbreaks  or conflict. However, the number of doses required to reach high efficacy may limit this application. There may also be a role in helping mitigate spread of drug resistance in the Mekong or for targeting forest workers , or for areas in which residual or outdoor transmission renders indoor residual spraying less effective, but further investigations are required.
Our modelling study provides evidence of potential use of an AIV in malaria elimination with characteristics similar to the most advanced vaccines: high initial efficacy and with limited duration of protection in an area with highly seasonal malaria transmission. Assumptions in regards to demographics, past history of malaria, transmission profile, vector species, and health systems strengths were not geographic specific. An important next step will be to understand which settings and countries these mass interventions strategies are feasible for and what range of settings with pockets of focal transmission. This would include an assessment of cost-effectiveness.