The following essay was contributed by public health researcher Jon Servello.

This essay makes extensive use of abbreviations. The list can be found here.

Malaria caused 627,000 deaths in 2020, largely among children under five years of age in sub-Saharan Africa. Interventions aimed at preventative treatment could save a considerable number of lives, bring about long-term economic benefits to countries with endemic malaria, and yield substantial benefits to humanity as a whole. 

After decades of varied success in vaccine development, a four-dose regimen of RTS,S/AS01 (brand name Mosquirix) showed promising initial results in preventing clinical disease from Plasmodium falciparum at 11 trial sites across west and central Africa, though this quickly declined to an average of 25.9% efficacy in the 6 to 12 week old age range, and 36.3% in the 5 to 17 month range. The European Medicines Agency licensed the vaccine in 2015, and RTS,S/AS01 is being rolled out via the Malaria Vaccine Implementation Program (MVIP) by the World Health Organization in Ghana, Malawi, and Kenya.

This article examines the potential benefits of the RTS,S/AS01 vaccine rollout as predicted by mathematical modeling exercises, what role the vaccine could play in the context of other malaria interventions, possible obstacles to full coverage, a risk-benefit analysis of rollout from an ethical perspective, and a final note on possible alternatives.

The role of mathematical modeling in malaria prevention

Mathematical modeling of malaria transmission dynamics has its origins in the early 20th century with Ronald Ross’s deterministic population model, was expanded on in the 1950s by George MacDonald with the addition of a variable for mosquitoes, and was used as a way of measuring the potential effects of indoor pesticide spraying. Such models were used in various forms until a renewed interest in elimination following the 2007 Gates Malaria forum spurred a need for more detailed models. 

In response to the desire to measure mortality and morbidity, as well as factor in the pathogenesis and heterogeneity of infection and immunity between individuals, stochastic individual-based models (IBMs) were developed. This led to the creation of the OpenMalaria, Imperial College London, EMOD DTK, and GSK models for simulating the effects of RTS,S vaccines.

Modeling based on hypothetical vaccine efficacy suggested a few key elements in the creation of a new vaccine for use in moderate to high transmission settings: 

• Initial efficacy is critical for interrupting transmission
• Providing partial protection to more people is better for reducing transmission than ‘hit-or-miss’ high and low efficacy in individuals 
• Partially efficacious vaccines would be best-deployed in settings with good coverage with existing tools. 

Perhaps most importantly, these models revealed that vaccination programs should be an important first step on the path to elimination, lest we risk an increase in clinical disease years later with diminishing population level immunity.

These models improved with real-world data from the RTS,S/AS01 phase 3 trial results, highlighting congruence with most simulations with a few notable exceptions; models assumed a standard of care for non-vaccine arms consistent with the health systems of the country setting, whereas cases of severe malaria were promptly and effectively managed in the real-world research setting, resulting in low placebo group mortality.

RTS,S in combination with other technologies

Using geographical data from the Malaria Atlas Project, further modeling the efficacy profile of RTS,S/AS01 suggests that RTS,S could have synergistic effects with mass drug administration (e.g. in the form of malaria chemoprevention), leading to a dramatic decrease in prevalence and, in some areas, a transmission interruption.

It is less clear how vaccination might interact with other technologies of interest to malaria elimination. One major component of predictive models for malaria interventions, for example, is the rate of exposure to infected mosquitoes. This is usually measured as entomologic inoculation rate (EIR), as shown in Smith et al.’s ‘ensemble’ models.

Genetic engineering technologies for propagating a particular set of genes (that result in infertility, for example) throughout a population have already been piloted in an effort to reduce the number of Aedes aegypti mosquitoes in the Florida Keys, United States. Earlier attempts in the Cayman Islands, Mexico, and Brazil saw a reduction of A. aegypti up to 95%, according to results of the manufacturing company Oxitec. 

It is not known whether these results are replicable for the Anopheles genus, the species of which are responsible for much of the malaria transmission in sub-Saharan Africa. Some early simulations have attempted to model the effects of gene-drive methods, finding that despite differences in setting and spatial heterogeneity of mosquito habitats, that small and repeated introductions of genetically-modified mosquitoes could substantially reduce the overall mosquitos across the whole geographic area. 

Gene-drive technologies, which further increase the heritability of engineered genes among a population, may provide a viable way of dramatically reducing or completely eliminating certain species of mosquito in the future. While some initial obstacles to gene-drive have been found, with some organisms developing “resistance” as the remaining compatible organisms continue to have viable offspring, there are still avenues to increasing effectiveness by combining multiple gene-drives and adjusting for fitness effects.

While there are currently no models combining the potential effects of genetic engineering or gene-drive with other existing malaria interventions, this avenue presents a particularly promising route to elimination, with even the earliest Ross-MacDonald models emphasizing the importance of mosquito ecology in disease transmission. 

The Metaculus Community has a forecast on the question of whether a gene drive will be used to eliminate a mosquito species by 2100. That question closed prior to the publication of this essay, and so I provide my forecast and my rationale below.

The below was first posted as a comment on the above question:

---

I'm optimistic about the use of the technology to dramatically reduce the population of a species in combination with other interventions, but only temporarily. It requires continual release of genetically modified mosquitoes for several generations, and once that stops, wild-type versions of the gene return in four years.

Some other methods to prevent that include multiple simultaneous gene-drives and "careful choice of ecological effects" (which I suspect means ecological intervention).

It may be easier to answer this question if we had a cost estimate for a (multiple) sustained release; long enough for the number of needed generations that would lead to extinction. Suppression seems to be affordable, but more than that seems to be prohibitive.

Right now, I'm answering this provisionally as a "no". The technology will certainly exist, but it won't happen without the money for a project with the explicit goal of eliminating a species.

In the same question's comments, @AngraMainyu asked:

What if it's a combination of a gene drive and other methods, such as some sort of insecticide that targets normal male mosquitoes, or some other methods?

To which I replied: Historically, malaria interventions have usually taken a multi-pronged approach. Given that gene-drive has encountered problems with "resistance", it seems likely that any gene-drive effort would also include currently existing interventions or some future iteration of them.

---

Cost and economic modeling

Analyses of vaccination costs have consistently revealed across a number of simulated settings that wastage and drop-off assumptions have huge implications. Despite the variance in the cost of delivery between countries, the greatest factor is the anticipated price of the vaccine.

Using data shared by the Ministries of Health in the MVIP pilot countries of Ghana, Malawi, and Kenya, Baral et al. found that a vaccine price of US$5 per dose (assuming the vaccine is donor-funded) would mean incremental financial costs range from US$1.70 (Kenya) to US$2.44 (Malawi) per dose, US$0.23 (Malawi) to US$0.71 (Kenya) per dose delivered (excluding procurement add-on costs), and US$11.50 (Ghana) to US$13.69 (Malawi) per fully vaccinated child.

Cost-effectiveness analyses (CEAs) of possible “scale-up” by Winskill et al. found that, when comparing RTS,S/AS01 against other potential interventions, that a combination of long-lasting insecticidal nets, indoor residual spraying, and seasonal malaria chemoprevention might save more DALYs than the vaccine. However, these predictions become increasingly non-linear as coverage increases and, as noted, RTS,S/AS01 may have synergistic effects with these existing interventions and even help set a path toward commitment to malaria elimination. 

Macroeconomic modeling of malaria vaccination programs has been somewhat neglected in the literature, with an emphasis on CEAs and budget impact analyses, with one notable exception provided by Yerushalmi et al., who estimate that full vaccination coverage (100%) in Ghana could increase GDP over 30 years by US$6.93 billion (in 2015 prices) above the baseline without vaccination, equivalent to an increase in annual GDP growth of 0.5%.

‘Ensemble modeling’

Smith et al. take an ‘ensemble modeling’ approach to determine the potential efficacy and rollout time for malaria vaccines depending on their cost, clinical properties, and deployment method. Of particular interest here are the comparisons between including the vaccine in the existing Expanded Program on Immunization (EPI) delivered at 1, 2, and 3 months old, EPI with catchup for children under 18 months, EPI with vaccination of school children aged 6 to 11 years, EPI with vaccination of school children at low coverage (≤ 50%), mass vaccination of three monthly rounds every 5 years, and mass vaccination with low coverage (≤50%) (Figure 1).

Figure 1. Smith et al.'s models of vaccine doses per program year by deployment strategy.

For context, the MVIP is currently being piloted as part of the existing EPI program, aimed at children at 5 to 6 months of age with no “catch-up” for older children. It is not yet known whether a different deployment method will be utilized across Africa if the results of the MVIP indicate a substantial benefit to the population. Smith et al.’s model predicts up to an additional 10 per 1000 malaria deaths would be prevented with the rollout of RTS,S alongside the existing EPI program (Figure 2).

Figure 2. Smith et al.'s predicted number of malaria-related deaths averted.

(A) EPI vaccination, with infectious mosquito bites per annum = 2,

(B) EPI vaccination, with infectious mosquito bites per annum = 20

Like Galactionova et al., the authors find that the addition to the EPI program would provide the most benefit in combination with other malaria prevention strategies, and the vaccine’s greatest strength is likely to be in the suppression of endemic malaria through a stable decrease in transmission.

The below was first posted as a comment on the above question:

---

My forecast reflects that I am hopeful for the results of the RTS,S rollout in the Malaria Vaccine Implementation Programme (despite some drop-off in efficacy in its phase 3 trial), as well as the possibility for even better vaccines with R21. The full evidence report from September 2021 showed really promising efficacies when combined with seasonal malaria chemoprevention:

"The incidence of clinical malaria, hospital admissions with severe malaria and deaths from malaria was 62.8% (95% CI 58.4, 66.8), 70.5% (95% CI: 41.9, 85.0) and 72.9% (95% CI: 2.91, 92.4) lower in the combined group than the SMC alone group. The incidence of these outcomes was 59.6% (95% CI: 54.7, 64.0), 70.6% (95% CI: 42.3, 85.0) and 75.3% (95% CI: 12.5, 93.0) lower in the combined group than the RTS,S/AS01 alone group."

As the WHO report in @Jgalt's comment shows, health services and their use have been hugely affected by the pandemic. Restrictions are likely to have had the same effect on TB, as well.

I'm also concerned with increasing vaccine hesitancy and how well pharmaceutical interventions will be received more generally. This is more difficult to quantify.

Gene-drive would be a viable technology for achieving the kind of massive drop posed in the question, but only with enough money for a continual, sustained release of genetically modified mosquitoes (perhaps of several types).

For those reasons I'm relatively low confidence that the 90% mortality reduction will be achieved within the next 8 years, but this might change if costs and other issues with gene-drive are solved very quickly, or the MVIP has outstanding data.

---

Challenges to rollout

Conducting an extensive systematic review of studies of malaria vaccination programs from 1947 to 2017, Dimala et al. found that the most persistent challenges were inadequate community engagement due to a lack of information about the vaccine, concerns about vaccine side effects, inefficient delivery of child vaccinations, and sub-optimal quality of health services. These results suggest that one of the most powerful tools to be used in a wide-scale vaccine rollout is transparency; in this case, the results of phase 2 and phase 3 trials, as well as the data currently being gathered as part of the MVIP, should be interpreted and disseminated in the public interest. Clear communication about safety data and the relative risk of vaccination in comparison to malaria would help to earn the trust of parents.

Risk-benefit analysis from the perspective of research ethics

Data from the MVIP will be invaluable in deciding whether to expand RTS,S/AS01 vaccination across Africa. Risk-benefit analyses will be essential in determining whether it is ethical to include as part of the EPI. It is also worth noting, though, that some important questions have already been raised during the MVIP. Many scientists, including Doshi, Van der Graaf et al., and others have expressed concern that, while RTS,S/AS01 has been approved by national and local authorities in the pilot countries, that the substantial research component of the MVIP program (including the collection of medical data for safety and efficacy studies), requires a number of other considerations be made according to CIOMS guidelines – the central one being that parents should be asked for a more explicit expression of informed consent for RTS,S alongside the existing EPI vaccinations. 

In light of the importance of public trust and the credibility of those organizations responsible for delivering vaccination programs already highlighted as potential obstacles, as well as historical evidence that any ethical controversy can derail national rollout programs (e.g. the HPV vaccine in India), these scientists suggest that programs that integrate public health interventions and research should involve the greater human rights and ethics community in addition to research ethics committees, as well as clear and transparent engagement with target populations.

Alternatives to RTS,S?

The R21 vaccine has also shown a very high efficacy in early trials in Burkina Faso at 74%, suggesting that even while the MVIP proceeds, a suite of options may become available when selecting a malaria vaccine in the coming years – all with different properties for efficacy, costs, and associated logistics. Data from the MVIP might well help to inform another rollout of future cheaper and more effective alternatives, as well as offer further opportunities for inclusivity and community engagement in wide-scale research. 

Abbreviations

CEA: Cost-effectiveness analysis

CIOMS: Council for International Organizations of Medical Sciences

DALY: Disability-adjusted life year

EIR: Entomologic inoculation rate; a measure of exposure to infectious mosquitos, often given as x IBPA

EMOD-DTK: Epidemiological Modeling Disease Transmission Kernel

EPI: Expanded Program on Immunization

GDP: Gross domestic product

GSK: GlaxoSmithKline

IBPA: Infectious mosquito bites per annum

MVIP: Malaria Vaccine Implementation Program

US$: United States dollars

WHO: World Health Organization

References

Baral, Ranju, Ann Levin, Chris Odero, Clint Pecenka, Collins Tabu, Evans Mwendo, George Bonsu, et al. "Costs of Continuing Rts, S/Aso1e Malaria Vaccination in the Three Malaria Vaccine Pilot Implementation Countries." PloS one 16, no. 1 (2021): e0244995.

Cook, Forest, James J Bull, Richard Gomulkiewicz, “Gene drive escape from resistance depends on mechanism and ecology”, bioRxiv 2021.08.30.458221; doi: https://doi.org/10.1101/2021.08.30.458221

Datoo, Mehreen S, Magloire H Natama, Athanase Somé, Ousmane Traoré, Toussaint Rouamba, Duncan Bellamy, Prisca Yameogo, et al. "Efficacy of a Low-Dose Candidate Malaria Vaccine, R21 in Adjuvant Matrix-M, with Seasonal Administration to Children in Burkina Faso: A Randomised Controlled Trial." The Lancet 397, no. 10287 (2021): 1809-18.

Dimala, Christian Akem, Belmond Tse Kika, Benjamin Momo Kadia, and Hannah Blencowe. "Current Challenges and Proposed Solutions to the Effective Implementation of the Rts, S/As01 Malaria Vaccine Program in Sub-Saharan Africa: A Systematic Review." PloS one 13, no. 12 (2018): e0209744.

Doshi, Peter. "Who’s Malaria Vaccine Study Represents a “Serious Breach of International Ethical Standards”." bmj 368 (2020).

Galactionova, Katya, Thomas A Smith, and Melissa A Penny. "Insights from Modelling Malaria Vaccines for Policy Decisions: The Focus on Rts, S." Malaria journal 20, no. 1 (2021): 1-8.

Mudur, Ganapati. "Human Papillomavirus Vaccine Project Stirs Controversy in India." British Medical Journal Publishing Group, 2010.

North, Ace R, Austin Burt, and H Charles J Godfray. "Modelling the Potential of Genetic Control of Malaria Mosquitoes at National Scale." BMC biology 17, no. 1 (2019): 1-12.

Penny, Melissa A, Robert Verity, Caitlin A Bever, Christophe Sauboin, Katya Galactionova, Stefan Flasche, Michael T White, et al. "Public Health Impact and Cost-Effectiveness of the Rts, S/As01 Malaria Vaccine: A Systematic Comparison of Predictions from Four Mathematical Models." The Lancet 387, no. 10016 (2016): 367-75.

RTS,S Clinical Trials Partnership. "Efficacy and Safety of Rts, S/As01 Malaria Vaccine with or without a Booster Dose in Infants and Children in Africa: Final Results of a Phase 3, Individually Randomised, Controlled Trial." The Lancet 386, no. 9988 (2015): 31-45.

Smith, Thomas, Amanda Ross, Nicolas Maire, Nakul Chitnis, Alain Studer, Diggory Hardy, Alan Brooks, Melissa Penny, and Marcel Tanner. "Ensemble Modeling of the Likely Public Health Impact of a Pre-Erythrocytic Malaria Vaccine." PLoS medicine 9, no. 1 (2012): e1001157.

Rieke van der Graaf, Ruth Macklin, Annette Rid, Anant Bhan, Eugenijus Gefenas, Dirceu Greco, David Haerry, et al. "Integrating Public Health Programs and Research after the Malaria Vaccine Implementation Program (Mvip): Recommendations for Next Steps." Vaccine 38, no. 45 (2020): 6975-78.

Winskill, Peter, Patrick G Walker, Richard E Cibulskis, and Azra C Ghani. "Prioritizing the Scale-up of Interventions for Malaria Control and Elimination." Malaria journal 18, no. 1 (2019): 1-11.

Yerushalmi, Erez, Priscillia Hunt, Stijn Hoorens, Christophe Sauboin, and Richard Smith. "Exploring the Use of a General Equilibrium Method to Assess the Value of a Malaria Vaccine: An Application to Ghana." MDM policy & practice 4, no. 2 (2019): 2381468319894345.