Models of Acquired Immunity to Malaria: A Review

■77

proving tolerance to sub-patent infections i.e. controlling malaria parasitaemia at low den-

sity and slower clearance of such infection. The results of their model show that the first

two mechanisms together account for the patterns of malaria by age group which match

with observed data in the African settings, whereas the latter was not needed to explain

the observed data. This model also suggests that immunity to symptomatic disease grows

faster with higher levels of infection in the population, can last for at least five years, and

increases with age-associated cumulative exposure. On the other hand, anti-parasite immu-

nity (which results in more rapid recovery from symptomatic or asymptomatic infections

to undetectable infections) develops later in life and can last for 20 years or more. Based

on the model, this aspect of immunity appears to better portray the pattern of parasite

prevalence and clinical disease by age than being exposure-dependent. By implicitly in-

corporating immunity functions, this model provides a framework to study time frames

over which clinical and parasitological immunity are likely to develop and dwindle, and

also the role of exposure and age on these functions. The concept of treating immunity as

a function of many parameters is different from the approach used by most deterministic

models which consider immune individuals as a separate class without considering the role

of immunity in disease progression.

An advanced way employed by some modelers is to model immunity as a function of

individual exposure history, where immunity memory can either grow by adding an expo-

sure value or decay by a factor with each iteration, based on whether or not an individual

got an infection [76], [109], [72], [102], [37]. These models represent immunity by a single

effector variable which regulates the course of the disease; such that infection scenarios are

different with respect to immune status accumulated over time. However there is no such

thing as completely immune and completely suceptible individual as some models would

assume [109], [110], [37]. In [109] for instance, Immunity ranged from 0 defined as “no

immunity” to 1 defined as “full immunity”. The ABM simulations of acquired immunity in

[37] show a strange spiraling behavior as time increases, which might have resulted from

the oversimplified assumption about NAI wherein an individual is classified as either com-

pletely immune or completely susceptible to infection, which is not biologically reasonable

[68], [36].

The classification of humans into discrete categories or just adding a value to immune

memory does not reveal the fact that malaria parasite populations can exist at varied den-

sities within the human hosts [30], [43], [126], [127], [128], [146], [160]. Thus, these

models give simplistic or no descriptions of the interaction between the population biol-

ogy of malaria and immune responses. However, immunity to malaria in an endemic area

is evident via both lower prevalence of detectable parasitemia with age and lower rates

of disease [36], [92]. Moreover, the density of asexual parasites is clearly associated with

disease risk, so diminished parasite counts as a result of immunity, obviously contribute to

reduced disease burden [34], [76], [78]. Aron [181] demonstrated how the underestimation

of the prevalence among older age groups, can also confound the statistics of prevalence,

rates of infection and apparent recovery. This suggests that an more realistic model will

have to follow a more detailed description in which the clinical symptoms and the im-

mune response are dependent on the density of parasite in an individual host rather than