Funded Grants

A complex systems approach to the immunological and epidemiological consequences of immunization for chronic infectious disea

The developed and developing world continue to be severely affected by the problem of chronic infectious disease. An estimated 40 million people are currently infected with HIV, with 5 million new infections and 3 million deaths in 2001. Tuberculosis and malaria also continue to exact a huge toll, estimated at 1.7 million and 1 million deaths respectively each year. In the developed world, the spread of hepatitis C virus also poses a threat to public health, with an estimated 2.7 million Americans chronically infected. The spread of these diseases can be contrasted with the success of the campaign to eradicate polio, which is now nearing completion. The contrast between the devastation of HIV and the near eradication of polio provides a strong incentive to push forward development of a vaccine for HIV.

Analysis of HIV vaccine studies in non-human primates suggest that within five to ten years it is likely that a human vaccine for HIV will emerge that produces at least some attenuation of disease in infected individuals. However, it appears unlikely that such a vaccine will resemble the more traditional vaccines we know to date. The HIV vaccines currently being developed are unable to prevent infection. Instead, they may simply reduce the severity of disease following infection. Therefore, a likely scenario is that one or more vaccines will emerge from phase III trials in the next 5-10 years, which stimulate the immune system and produce significant reductions in the concentration of virus. In an ideal study, these individuals would be followed for a decade or more to ascertain their long-term survival following infection, their viral load and their risk of transmitting the virus with time. However, given the urgency of delivering a vaccine of any sort, it seems unlikely that both scientists and the community will be prepared to wait the many years required to acquire this empirical data. In the absence of this data, how will it be possible to estimate the minimal vaccine effectiveness required to significantly reduce HIV transmission? Indeed, it is not even clear what measures of vaccine effectiveness are important for this estimation. If two competing vaccines show slightly different effects for example one decreasing the concentration of virus and the other more significantly slowing the rate of disease progression and delaying the time to AIDS - how will it be possible to estimate the relative long term benefits of the two?

In order to assess the likely effectiveness of a vaccine it is necessary to understand both the effects of the vaccine on host immune responses and on viral evolution. The immune system employs a variety of mechanisms to fight infection, including 'helper' and 'killer' T cells, as well as B cells which produce antibody. The killer T cell response is thought to be most important in fighting HIV and most other chronic infections. None of the vaccines we presently use (against polio, measles etc) relies on these killer T cell responses. Most current vaccines against common illnesses depend on antibody to exert their protective effects. More importantly, our current vaccines are all against 'acute infections' - ones in which you get infected, become ill (or die), then completely eliminate the virus from you body and remain immune to re-infection for life. That is, exposure to these infectious agents - either through natural infection or vaccination - always lead to eradication of the virus and long-term immunity. This is in complete contrast to HIV and other chronic infectious diseases. In HIV it is unclear that anyone has ever eliminated the virus through natural immunity. It also seems that the immune response to HIV does not prevent re-infection some people may be infected with two different strains of HIV at different times. However, this does not mean that immune responses to HIV are useless. Some people infected with HIV manage to survive for decades without developing AIDS, probably due to a strong immune response. Similarly, as described above, vaccines in animals seem capable of delaying AIDS for substantial periods.

Our current understanding of how the immune system controls HIV is very limited. We do understand, however, that the viral-immune interactions are very complex. The immune response targets small regions of the virus, and these regions can mutate and change to avoid this immune recognition. In fact, over the course of infection the immune system targets seem to constantly shift to new regions of the virus as the virus 'escapes' immune recognition of earlier sites. Some think that this constant 'race' between the immune system and virus - the virus escaping the immune response and the immune system fording new targets - eventually 'exhausts' the immune response. Since animal models suggest that vaccination can't prevent infection, perhaps the next best thing would be to slow down these cycles of 'escape'? However, in order to achieve this we must first understand the details of this immune-viral interplay.

Developing a realistic model of immune responses to a virus can be done using techniques from complex systems theory and a computer simulation technique called 'lazy evaluation'. Killer T cells 'recognize' the virus through T cell receptor molecules on their surface. The body contains somewhere around a billion different T cells. These cells recognize different parts of the virus by the binding of different molecular shapes on the T cells to the molecules of the virus. These molecular interactions can be simulated on a computer by representing strings of amino acids (the building blocks of proteins) as strings of numbers. When the molecules (represented as number strings) on a T cell complement the molecules of the virus (also represented as strings), then the T cells can divide and kill the virus. In infection, viruses can mutate by changing the amino acid sequences in their proteins. Similarly, the computer model allows for changes in the number strings representing the virus. It is extremely difficult to monitor all these cells in an infected individual. We are currently developing computer simulations that keep track of the number, specificity (region of the virus recognized) and activity of thousands of different T cells. Since HIV is also able to mutate rapidly, the model must also keep track of many different viral mutants. Such models make it possible to ask "what if?" questions about different approaches to vaccination, without having to resort to large scale trials in animals or humans. Different vaccination strategies that are predicted by the model to be likely candidates for success can then be trialed and the results fed back into the model to further optimize the analysis.

The models just described help us predict the effects of a vaccine within an infected individual, but what about the effects of a vaccine within the population? An ideal vaccine not only has to allow infected individuals to live longer, it also has to reduce the transmission of the virus and therefore prevent new infections. To predict how well a vaccine might reduce the spread of HIV requires a different type of model - an epidemiological model - which tracks spread of the virus within the population. However, the dynamics of this model are very much dictated by the model of immuneviral interactions within individuals. For example, if vaccination reduces the concentration of virus in the blood of an individual, it will make it less likely that the virus will be transmitted. On the other hand, if the virus 'escapes' the immune response to vaccination in one person and this 'escape mutant' virus is transmitted to a second person, then the vaccine will have no protective effect (since the virus has already 'escaped' its response).

The aim of any vaccination campaign is to maximize the survival of each infected
individual and to minimize the number of new infections. This requires the combination of the immunological model of what happens within an individual, and the epidemiological model of what happens in a population. Optimizing all of these parameters simultaneously requires consideration of the multiple levels on which the virus operates - molecular, cellular, within an individual and within a population. Although we currently have some computer models which allow us to analyze the effects of virus on one of these levels or another, we do not have any models which incorporate the complexities of human-viral interaction. Without sophisticated models that incorporate all these aspects of a vaccine's effectiveness to help us predict likely outcomes of vaccination, we may either have to wait years to acquire the data through experimental observation - or risk mass administration of a vaccine without a full understanding of the likely outcome.

Approaches to treatment of bacterial and viral infections in the past have often been
based on little understanding of the long-term implications. The rise of antibiotic resistant bacteria and multi-drug resistant HIV provide salutary examples of how the usefulness of different weapons used in the fight against infectious disease can quickly be undermined. By the use of complex systems analysis of human-viral interactions it is hoped that we can avoid a similar situation occurring with future vaccines for HIV and other chronic infectious diseases.