There is an increasing trend to investigate biomedical systems through the prism of Complexity Science, (as highlighted, for instance, by recent funding calls). This is, in part, motivated by the significant translational potential of models developed in this context.
When considering the immune response to HIV infection, complexity is a direct consequence of the diversity and multitude of cells involved, their mobility patterns, and the various interactions between these cells and the virus. Recent studies have, however, highlighted another layer of complexity, in the sense that local properties of the immune system influence the overall disease progression. Early infection in the gastrointestinal tract, in particular, is crucial [1].
In this paper, we report on a large-scale agent-based model, with explicit implementation of lymph nodes. This structure is tested on a uniform network of nodes, as well as on a network including local properties, to investigate changes in the dynamics of overall disease progression. This model also serves as a basis for discussion on the potential of similar approaches for other biomedical systems.
Given that the immune system is characterised by emergent properties, it is particularly suited to an agent-based approach, with agents uniquely identified with cells. Several such models have been developed to address questions of interest, (see e.g. [2, 3, 4]). In their current form, however, these do not easily permit investigation of the impact of local properties on behaviour of the system as a whole.
These properties may lead given areas to exhibit very distinct patterns: this includes the gastrointestinal tract. Of major importance is the role of this tract in terms of overall immune population: it harbours the majority of the body's lymphocytes, where for instance blood only accounts for a few percent of these [5]. Even more importantly, these cells are in close proximity to the external environment and are, therefore, constantly exposed to countless antigens. This results in two crucial properties: more than 90% of these lymphocytes have a memory phenotype, and the proportion of activated cells is significantly higher [6, 7]. With HIV primarily targeting CD4+ cells, (activated lymphocytes), these factors mean that there is, typically, a large population of targets, (i.e. potential host cells), for the virus to infect and, therefore, a massive infection in the tract even in the early stages of the infection. The immune response in this specific area is, therefore, also very active. Recently published experimental results show:
-
A very rapid and very significant decline in CD4+ counts, exceeding 25% after four weeks of infection [1].
-
Significant levels of infection and destruction observed even within days of infection for memory CD4+ cells [8].
-
An increased cell proliferation in response to infection. The cell proliferation marker was found on 80% of intestinal CD4+ cells four weeks after infection, as opposed to on less than 10% in healthy patients [1].
Due to this local but substantial depletion of immune cells, the overall cell population is also severely reduced, and this imposes significant pressure on the immune system in terms of memory pool maintenance [9]. It also damages lymphoid tissue architecture, and this hinders the ability to support normal lymphocyte homeostasis and antigen presentation.
Early infection in the gastrointestinal tract has, therefore, become an essential of research against HIV. Exact implications of GI tract infection remain largely unclear, but some interesting progress is being made. At the molecular level, it has been shown that preferential targeting of gut-associated CD4+ cells may be due to interactions between viral glycoprotein gp120 and integrin α4β7, which is specific to these cells [10].
At a higher level, restoration of cell populations after the acute phase is also under investigation. Observations highlight a delayed and incomplete restoration of cell populations in chronically infected patients, even for those receiving highly active antiretroviral therapy, (HAART), for more than five years [1]. This standard therapy leads to restoration of cell levels in peripheral blood, but not in the tissue considered. A similar therapy, however, if initiated during primary infection, is effective in restoring cell populations. In this case, restoration is a consequence of cell recirculation and increased homing from the periphery of the tract, rather than of local cell proliferation.
To evaluate the potential for such treatment policies, it is crucial to understand, in a first instance, how these locally-altered system properties influence the overall infection progression, a crucial objective of the model presented here. Given the importance of the GI tract in the overall progression and its response to the infection itself and to proposed therapies, it is essential to consider the tract within the whole system, rather than as an independent subject. Using a single-node environment would permit the investigation of some aspects of GI tract dynamics, but would not provide any means to study their impact on the overall immune system. A network representation of the lymph network is, therefore, proposed. Tests are performed on large clusters, suitable for massively multi-agent simulations. Here, we reach an agent count in excess of one billion.