JAMES MALM AFRICAN INSTITUTE FOR MATHEMATICAL SCIENCES AIMS ESSAY PRESENTATION 1
Long term Dynamics of HIV disease progression with a non-sterilizing vaccine 0.1 Introduction Vaccines have been used for nearly three hundred (300) years to prevent infectious diseases. The vaccine, vaccinia has been used to eradicate small-pox. The near-eradication of poliomyelitis can be attributed to effective vaccination strategies. However, there are still some infectious diseases for which no vaccine has been developed. To date, no vaccine has been shown to protect humans from HIV infection. The only successful HIV vaccines are the CTL vaccines that stimulate the CTL immune response. There are two (2) types of virus namely: The wild-type virus The wilder-type virus (escape mutant virus) The CTL vaccines can control the wild-type virus. The CTL vaccines CANNOT control the wilder-type virus (escape mutant virus). Recent studies in animal models suggest that it is possible to produce an HIV vaccine. This HIV vaccine comes with a good news and a bad news. The bad news is that the vaccine ALLOWS INFECTION. The other bad news is that the vaccine has NO EFFECT on the wilder-type virus (escape mutant virus). the good news is that the vaccine can reduce the viral load of those infected with the wild-type virus. the other good news is that the vaccine can slow the decline of the CD4 cells in people infected with the wild-type virus. This vaccine is called a DISEASE MODIFYING VACCINE. If the results seen in animals models are replicated in human infection, then it seems that in a short to medium term a human HIV vaccine will emerge. 2
This vaccine will significantly increase survival of infected individuals. This vaccine will also decrease HIV mortality. Such vaccines will clearly benefit infected individuals by delaying the disease progression. However, the long-term impact of the vaccines on the epidemic as a whole is UNCERTAIN. If the vaccine reduces viral load patients may experience a longer latent period of infection. Patients still ultimately progress through the same advanced stages of the disease as seen in natural infection. In addition, there is concern that optimism over the effects of the vaccine may lead to an increase in sexual behaviour and this might increase HIV transmission as observed following the introduction of the antiretroviral therapy. 0.2 Objective This research project assesses the long-term impact of a disease modifying HIV vaccine in southern Africa. We will achieve this objective by assessing the: impact of disease modifying vaccines on HIV mortality and incidence in southern Africa. settings in which disease modifying vaccines will be effective. the features (parameters) of the vaccine which are important on mortality and incidence. 3
Chapter 1 Model Development We develop mathematical models in order to assess the long-term impact of a disease modifying vaccine. We adapt a previously published age-structured epidemic model for a disease modifying vaccine [Davenport, 2004]. The model has 12 age groups starting from 15 19, 20 24, 30 34,...,70+. The model considers: how age and sex may affect virus transmission. changes in viral load and disease progression affect mortality and incidence. age specific partner change, sexual mixing. 1.1 Changes made to the Davenport Model Collaborated with the Actuarial Science department to obtain data specific to southern Africa on: HIV mortality rates. natural mortality rates. sexual mixing matrix. age specific partnership change. Please refer to Figure A.6 on page 39 for the model assumptions and description. The model is structured into two major compartments namely the: susceptible. infected. Let us talk about the susceptible compartment. In the susceptible compartment we have the: 1. unvaccinated susceptible. 2. vaccinated susceptible. The infected compartment consists of: 1. unvaccinated infected with the wild-type virus. 2. vaccinated infected with the wild-type virus. 3. (un)vaccinated infected with the wilder-type (escape mutant virus). 4
1.2 Model Assumptions Some of the model assumptions are: vaccine has a finite duration. vaccine has no effect on the wilder-type (escape-mutant) virus. viral load increases with time. the effects of vaccination is lost once the viral load increases to the level observed in natural infection. individuals infected with the wild-type virus can be also infected with the wilder-type (escape mutant) virus (SUPER-INFECTION). Using the model s assumptions we can write some elegant differential equations to show the transitions among the various classes of the model. 5
Chapter 2 Model Simulation 2.1 Impact of disease modifying vaccines on HIV mortality and incidence in southern Africa In order to assess the impact of disease modifying vaccines in southern Africa, we assessed the impact of two vaccines on HIV mortality and incidence. a vaccine that reduces viral load. and a vaccine that slows disease progression. we assumed an initial prevalence rate of 10% of the sexually active individuals. We carried out the study for a simulation time of 25 years for the vaccine that reduces viral load and the vaccine that slows disease progression. See Figures 3.1 3.6 on pages 15 17. Our results show that a vaccine that reduces viral load is a better option than the vaccine that will slow disease progression. The vaccine that reduces viral load: increased the survival of HIV infected individuals. decreased HIV mortality. The vaccine that reduces viral load was able to prevent about 45% of HIV mortality after 25years. 2.2 Settings in which a disease modifying vaccine will be effective We studied the settings in which the disease modifying vaccines will be effective. We investigated the impact of the two vaccines in different simulated populations. For each of the population we allowed the epidemic to develop without vaccination until the prevalence reached 1%, 2%, 4%, 5%, 8%, 10%, 16% and 20%. Our results reveal that: for any given growth rate, the lower the initial prevalence, the larger the impact of the vaccine in: increasing the survival of HIV infected individuals. decreasing HIV mortality. Again the vaccine that reduces the viral load had more impact on HIV mortality and incidence than the vaccine that reduces viral load. See Figures 3.8 3.9. 6
Next we studied a vaccination scenario in which there was a 1 log 10 copies ml 1 drop in viral load following vaccination for the different simulated populations. The was a reduction in HIV mortality for all the populations we considered. We believe that there is low risk that a vaccine that reduces viral load by at least 1 log 10 copies ml 1 will increase HIV associated mortality. 2.3 Features (parameters) of the vaccine which are important on mortality and incidence. We studied how the parameters of our model affect HIV mortality and incidence to know: how changes in parameters affect the model. which parameters are significant in affecting HIV mortality and incidence. which parameters are insignificant in affecting HIV mortality and incidence. We used the Plackett-Burman sensitivity analysis technique. This technique involves: simultaneous investigation of all parameters. calculation of two-way interaction effects. requires 2d scenarios (d = design size). for n parameters d > n and mod(d, 4) = 0. Our model has 7 parameters ie n = 7. The parameters are shown on Table 4.3 on page 24. we choose a design size d = 8. we have 2d = 2 8= 16 scenarios. for paired interaction we have C(7, 2) = 21 scenarios. Sensitivity analysis results show that: viral load reduction has the major influence on HIV mortality and incidence. Risky behaviour (R) has a major effect on HIV mortality and incidence. Loss of vaccine effectiveness (ω) has a major influence on HIV mortality and incidence. Rate of vaccination (V ) has a major influence on HIV mortality and incidence Transmission of escape variants (H E ), viral escape (ɛ) and progression rate of vaccinated individuals (p v ) had no significant effect on HIV mortality and incidence. 7
Chapter 3 Summary and conclusion In this project we assessed the long-term dynamics of a disease modifying vaccine in southern Africa. We used an age-structured HIV epidemic spread model that incorporates intra-hosts and population dynamics to study the effects of a disease modifying HIV vaccine. we assessed the impact of two vaccines on HIV mortality and incidence: - a vaccine that reduces viral load. - and a vaccine that slows disease progression. Our results showed that a vaccine that reduces viral load can reduce HIV mortality and increase survival of infected individuals. We studied the settings in which the disease modifying vaccines will be effective. We investigated the impact of the two vaccines in different simulated populations. Results show that the growth rate, the initial prevalence and the duration of clinical trials are important factors to consider if we want to assess the impact of the vaccine in different populations. A disease modifying vaccine will have a smaller impact when introduced into a late stage, high prevalence epidemic in areas of sub-saharan Africa. A disease modifying vaccine will have a greater impact when introduced into relatively early stage epidemics. Our results also show that a drop in viral load of about 1 log 10 copies ml 1 is sufficient to produce long-term reductions in HIV incidence and mortality. We studied the features of the vaccine that affect HIV mortality and incidence by carrying out a sensitivity analysis. Sensitivity analysis results suggest that to predict the likely outcome of a disease modifying vaccine. It is important to measure the reduction in viral load in vaccinated individuals relative to unvaccinated individuals. Both vaccination rate and rate of loss of vaccine protection had major influence on outcomes. Increase risk behaviour had a strong influence on outcomes as predicted in previous models. 8
Surprisingly according to our model the extent to which vaccine slows disease progression and rate of emergence and transmission of immunological escape virus had insignificant effects on HIV mortality and incidence. 3.1 Discussions Our model suggests that the success of any human vaccine trial should not be judged solely on shortterm measures of how many infections were prevented in the vaccinated group. It is likely that disease modifying vaccines would lead to a short-term increase in HIV incidence if they lead to increased risky behaviour 3.2 Recommendations It is important to link effective behavioural intervention strategies with vaccination programs Vaccine trials should monitor viral loads of infected individuals to quantify any reduction in viral load of vaccinated individuals. Model can be extended to include the impact of high risky groups like intravenous drug users, commercial sex workers or other core groups within a population. 3.3 Model Application Model can be used as a health-policy tool to predict the future impact of disease modifying vaccines in controlling the HIV epidemic in southern Africa and other developing countries with similar epidemic profiles. 9