Group Title: Malaria Journal 2008, 7:67
Title: Clinically immune hosts as a refuge for drug-sensitive malaria parasites
Full Citation
Permanent Link:
 Material Information
Title: Clinically immune hosts as a refuge for drug-sensitive malaria parasites
Series Title: Malaria Journal 2008, 7:67
Physical Description: Archival
Creator: Klein EY
Smith DL
Boni MF
Laxminarayan R
Publication Date: 39563
 Record Information
Bibliographic ID: UF00100197
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: Open Access:


This item has the following downloads:

clinically ( PDF )

Full Text

Malaria Journal BioIed Cen


Clinically immune hosts as a refuge for drug-sensitive malaria
Eili Y Klein1,2, David L Smith3, Maciej F Bonil,2,4 and
Ramanan Laxminarayan* 1,4

Address: 'Resources for the Future, 1616 P St., NW, Washington, DC 20036, USA, 2Department of Ecology and Evolutionary Biology, Princeton
University, Princeton, NJ 08544, USA, 3Department of Zoology, University of Florida, Gainesville, FL 32611, USA and 4Princeton Environmental
Institute, Princeton University, Princeton, NJ 08544, USA
Email: Eili Y Klein; David L Smith; Maciej F Boni;
Ramanan Laxminarayan*
* Corresponding author

Published: 25 April 2008 Received: 28 January 2008
Malaria journal 2008, 7:67 doi: 10. 186/1475-2875-7-67 Accepted: 25 April 2008
This article is available from:
2008 Klein et al; licensee BioMed Central Ltd.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Background: Mutations in Plasmodium falciparum that confer resistance to first-line antimalarial
drugs have spread throughout the world from a few independent foci, all located in areas that were
likely characterized by low or unstable malaria transmission. One of the striking differences
between areas of low or unstable malaria transmission and hyperendemic areas is the difference in
the size of the population of immune individuals. However, epidemiological models of malaria
transmission have generally ignored the role of immune individuals in transmission, assuming that
they do not affect the fitness of the parasite. This model reconsiders the role of immunity in the
dynamics of malaria transmission and its impact on the evolution of antimalarial drug resistance
under the assumption that immune individuals are infectious.
Methods: The model is constructed as a two-stage susceptible-infected-susceptible (SIS) model of
malaria transmission that assumes that individuals build up clinical immunity over a period of years.
This immunity reduces the frequency and severity of clinical symptoms, and thus their use of drugs.
It also reduces an individual's level of infectiousness, but does not impact the likelihood of becoming
Results: Simulations found that with the introduction of resistance into a population, clinical
immunity can significantly alter the fitness of the resistant parasite, and thereby impact the ability
of the resistant parasite to spread from an initial host by reducing the effective reproductive
number of the resistant parasite as transmission intensity increases. At high transmission levels,
despite a higher basic reproductive number, R0, the effective reproductive number of the resistant
parasite may fall below the reproductive number of the sensitive parasite.
Conclusion: These results suggest that high-levels of clinical immunity create a natural ecological
refuge for drug-sensitive parasites. This provides an epidemiological rationale for historical patterns
of resistance emergence and suggests that future outbreaks of resistance are more likely to occur
in low- or unstable-transmission settings. This finding has implications for the design of drug policies
and the formulation of malaria control strategies, especially those that lower malaria transmission

Page 1 of 9
(page number not for citation purposes)

Malaria is the leading cause of death in children under five
in sub-Saharan Africa [ 1]. Prompt treatment with effective
antimalarial drugs could prevent much of the morbidity
and mortality associated with clinical malaria, but the
evolution of resistance has diminished the therapeutic
efficacy of two previous first-line antimalarials, chloro-
quine (CQ) and sulphadoxine-pyrimethamine (SP). His-
torically, it has been suggested that resistance to both CQ
and SP emerged from a limited number of de novo selec-
tion events in areas of low or unstable transmission [2,3].
Genetic evidence suggests that most CQ-resistant parasites
in the world today are descended from one of four
founder events [4,5] that occurred in Southeast Asia,
South America and Papua New Guinea, and then spread
to other regions, including sub-Saharan Africa. Evidence
also suggests that resistance to SP originated from only a
few foci [6-8]. Several hypotheses about the de novo muta-
tion rate and the selection pressure resistant parasites face
and their relation to transmission intensity have been pro-
posed to explain why CQ resistance originated in what is
presumed to be low or unstable transmission areas out-
side of sub-Saharan Africa, including (a) a lower fre-
quency of resistant alleles in higher transmission areas
because of within-host competition [9]; (b) less drug
treatment (per parasite) in higher transmission areas
[2,9,10]; and (c) a lower frequency of selfing in higher
transmission areas, which increases the probability that
multilocus resistant genotypes will be broken up by the
action of Mendelian segregation [9,11,12]. An additional
explanation is that in high transmission areas, where
immunity is better developed, mutant parasites are less
likely to survive a host immune response [10,13].

Although the evolution of resistant parasites within a sin-
gle host is a significant risk factor for the emergence of
resistance within a population, the effectiveness of an
antimalarial drug is affected only if these resistant para-
sites spread, a process that is driven both by the overall
rate of transmission and by the relative fitness of drug-
resistant and drug-sensitive parasites. Parasite fitness is
related to the population treatment rate and to the biolog-
ical cost of resistance. Because of differences in vector
ecology and biting preferences, transmission intensities
vary from less than one infectious bite per decade to more
than 1,000 per year [14], and the basic reproductive
number (R0) can exceed 3,000 [15]. Higher transmission
intensity is associated with a higher level of clinical immu-
nity to malaria reduced frequency and severity of clinical
symptoms in older children and adults which results in
a reduction in the need for antimalarial drugs [16-19].
Immunity to malaria has consequences for transmission
as well; blood-stage immunity reduces asexual parasite
and gametocyte densities in older children and adults
[20], and transmission-blocking immunity can block

development of the parasite in the mosquito [21]. Impor-
tantly, despite significant reductions in clinical symptoms
and infectiousness, older individuals still become infected
and remain infectious to mosquitoes, albeit at relatively
lower levels, even in holoendemic areas [22-24].

Even though immunity affects both the transmission
dynamics and treatment rates, epidemiological models
have generally assumed that immune individuals are not
infectious to mosquitoes, and thus do not contribute to
parasite fitness [20,25-27]. Consequently, an epidemio-
logical model for the spread of resistance found no differ-
ence in the ability of resistant parasites to spread at
different transmission rates [27], or why resistance to CQ
emerged outside of sub-Saharan Africa.

This paper reconsiders the role of immunity in the trans-
mission dynamics of malaria and its effect on the evolu-
tion of antimalarial drug resistance under the assumption
that immune individuals are infectious. Using a novel epi-
demiological model with two immune stages, nonim-
mune and clinically immune, the role of immunity in the
emergence of resistance is reexamined based on quantita-
tive effects associated with clinical immunity. Individuals
in the clinically immune stage have lower transmission,
lower incidence of clinical disease, and consequently a
lower rate of drug treatment. Based on the defined rela-
tionship between transmission intensity, immunity, and
clinical malaria, the model is used to explore the relation-
ship between vector ecology, human epidemiology, and
the ability of a resistant parasite to spread.

Individuals living in areas of endemic P. falciparum trans-
mission develop immunity to malaria with age [28] and
exposure [29,30]; immunity is manifest as a decline in
parasite densities (both trophozoites and gametocytes) in
the blood [20], a lower probability of transmission from
humans to mosquitoes [22,23], and a decline in the fre-
quency and severity of clinical malaria [16-19], though lit-
tle decline in the probability of becoming infected
[24,31]. Thus, it is assumed that individuals develop a
form of clinical immunity over time in which they are less
likely to infect mosquitoes or manifest clinical symptoms
but are no less likely to become infected.

The model is based on earlier models developed for the
Garki Project [20], in which individuals acquire immunity
after being infected for a period of time. However, the
number of infected classes was simplified and the
assumption that individuals develop full transmission-
blocking immunity was relaxed. As with the Garki model,
clinical immunity is incorporated as a second immune
stage in a susceptible-infected-susceptible (SIS) model.
Though, in this model clinically immune individuals

Page 2 of 9
(page number not for citation purposes)

Malaria Joumnal 2008, 7:67

remain infectious to mosquitoes and infect mosquitoes
with lower probability; also, a smaller fraction of new
infections progress to clinical malaria. Clinical episodes of
malaria are important for the evolution of resistance
because individuals who develop clinical symptoms are
more likely to seek treatment; thus an increasing fre-
quency of clinical episodes is associated with increasing
drug pressure. Consequently, in this model clinically
immune individuals represent a refuge for drug-sensitive
pathogens because of lower treatment rates. The evolution
of resistance is incorporated by assuming that individuals
can be infected by either resistant or sensitive parasites.

Population dynamics
It is assumed that both nonimmune and clinically
immune individuals can be susceptible, infected with sen-
sitive parasites, or infected with resistant parasites. The
human population density in each state is denoted S, I,,i,
I,, where the subscripts w and x denote infections with
drug-sensitive wild-type and resistant phenotypes, respec-
tively, and the i subscript denotes the immune stage. The
population size is normalized to one, and the population
birthrate B is set equal to the per capital death rate of the
human population, p, so that the total population size
stays constant.

Model notation follows Macdonald [32] and Smith and
McKenzie [33]; m denotes the number of mosquitoes per
human and a the human feeding rate (the number of bites
on humans per mosquito per day). The instantaneous
death rate is g (e is the probability of a mosquito surviv-
ing one day), and n is the number of days required for
sporogony. Vectorial capacity (V), the number of infec-
tious bites by a mosquito over its lifetime, is then given by
the formula V = ma2e-gl/g.

The fraction P of bites on humans that infect a mosquito
depends on the differing transmission intensities of the
two-stages, cI and c2, and the number of humans in each
stage; thus P = cl(Iwl + I,1) + c2(Iw2 + Ix2). The sporozoite
rate, or the fraction of infectious mosquitoes, is aPe-g /(g +
aP). The entomological inoculation rate (EIR), the
number of infectious bites per person per day, is calcu-
lated as the product of the human biting rate (ma) and the
sporozoite rate (aPe -'/(g + aP)). The force of infection, or
happenings rate (h), is bEIR, where b, the infectivity rate,
measures the fraction of bites in humans that produce a
patent infection. It follows that h=(bVP)/(1 + sP), where s
= a/g is called the stability index, the number of bites on a
human per vector per lifetime. The fraction of infections
that are drug sensitive is F, = (c1Ik1 + c2IJw)/P and the frac-
tion that are drug resistant is F = (cl4I + c2l2)/P. Happen-
ings rates for drug-sensitive and drug-resistant infections
are h, = Fwh and h = Fxh, respectively.

Immunity acquisition and parasite clearance
Clinical immunity is assumed to develop in infected indi-
viduals after ten years. Once individuals gain immunity,
protection is retained through biting and is lost at a faster
rate (y) than it is gained (6). The values are based on a sig-
nificant number of age-prevalence studies suggesting that
children acquire immunity after approximately five to 10
years [29,30,34], and additional studies that suggest a
strong role of biting in maintaining immunity, and a loss
of immunity that occurs after exposure to infection is
eliminated [35]. Because not all infections result in fever
and other associated symptoms, it is assumed that clinical
symptoms arise in infected individuals at a rate 7i and that
a fraction, are treated and cleared of parasites. Thus, exist-
ing infections are cleared by drugs at the rate pi = fi,. It is
further assumed that a fraction of new infections in sus-
ceptible individuals develop clinical symptoms and are
treated with drugs and cleared immediately prior to the
development of gametocytes, thus precluding the possi-
bility of transmission. In these individuals it is as if the
infection never occurred.

Mutations conferring resistance to antimalarial drugs are
likely to be disadvantageous to the parasite. For example,
resistance to CQ in P. falciparum has been shown to have
a fitness cost of approximately 25 percent in vitro [36] and
5 percent in vivo [37,38]. Although a fitness cost could
conceivably occur at any or all stages of the parasite life-
cycle, where it is implemented in this model is not partic-
ularly important for the transmission dynamics (though it
could be important in other models), so it is assumed that
resistant infections are cleared at a faster rate and therefore
transmit for a relatively shorter period. Thus, infections
clear naturally at rate r, when an individual is infected
with a drug-sensitive phenotype and r for drug-resistant

Based on the above assumptions, model dynamics are
described by a simple set of coupled ordinary differential

Si = B + YS2 + I1(p, + r.) + Ir( Sl(h(1 1,) + h, + )

Iiw = Slh(1 41) iwl(pi + rw + 0 + P)

ix Sbli _I'1(r +0+P)

S2 = Iw2(2 + r) + x2 S2(hw(1 2) + h + y + P)

iw2 = S2hw(1 2) + Iw1 w2(P2 + r + +)

Page 3 of 9
(page number not for citation purposes)

Malaria Journal 2008, 7:67

Figure I
SIS two-stage model. Susceptible individuals (S) become infected (I) with wild-type infections at the rate h,(I-i) and resist-
ant infections at the rate hx, where h, is the happenings rate (see text) and is the rate at which new infections result in clini-
cally manifested symptoms that are treated and resolved prior to the formation of gametocytes. Infected individuals naturally
clear resistant infections at rate rxi, and they clear wild-type infections at the rate p,+r,,, where p, is the rate of drug treatment.
Infected individuals acquire semi-immunity at rate 0and lose immunity at rate yif they are no longer infected. Individuals die
from all states at a background rate of gt and are born at rate B as nonimmune susceptibles (process not shown).

Ix2 = S '1 + IxO Ix2(r + i) (6)

A diagram of the model is found in Figure 1.

The principal assumptions for the analysis were that
immune individuals are less infectious to mosquitoes (c2
treated (p2 < p1 and 2 < 1), and that parasites face a resist-
ance-related fitness cost (rx > rJ). While recent evidence
has suggested that individuals with clinical-immunity
may recover at faster rates [39], it was assumed that recov-
ery from infection was independent of immune status in
the base model. Based on these assumptions, the equilib-
rium of the system was determined without resistance and
then the resulting spread of resistance was simulated
across a range of transmission rates (from less than one to
more than 600 infectious bites per person per year) using
a set of baseline assumptions (Table 1).

At equilibrium without resistance, the results were con-
sistent with other models of transmission. As vectorial
capacity increases, the proportion of the population that
is nonimmune, and thus subject to frequent attacks,
increases sharply and then slowly decreases, while the

proportion of immune individuals increases steadily until
it plateaus at a fairly high level (Figure 2). As expected
[24,31], the infected proportion of the population nears
100 percent at high transmission levels, though a signifi-
cant portion of infected individuals are clinically
immune. In addition, the results suggest that clinical epi-
sodes of malaria a proxy for hospital admissions for
malaria as well as severe malaria plateau at intermediate
levels of transmission, which is consistent with other
research [30].

Because symptomatic individuals are more likely to use
antimalarial drugs, the number of individuals experienc-
ing a clinical episode affects the transmission dynamics by
changing the drug pressure facing the parasite. Individuals
that acquire clinical immunity, have reduced rates of clin-
ical episodes [16-19], which reduces the rate of drug use.
Though individuals will continue to receive antimalarial
treatments throughout their lives, these treatments occur
less frequently as immunity increases, and often can be
unconnected to the peaks of parasitemia [3]. This reduc-
tion in the treatment rate reduces the resistance selection
pressure, providing a refuge for drug-sensitive parasites. In
previous epidemiological models of transmission, these
immune individuals were not assumed to have a signifi-
cant qualitative effect on the dynamics of the system,

Page 4 of 9
(page number not for citation purposes)

Malaria Journal 2008, 7:67

Malaria Journal 2008, 7:67

Table I: Baseline Parameter Values

Acquisition of clinical immunity
Loss of clinical immunity
Fraction of new infections that are treated and cleared

Rate clinical symptoms arise (o,) times Fraction treated (f) equals rate existing infections are cleared by drugs (p)

Disease induced death rate
Human feeding rate
Infectivity rate
Mosquito death rate
Number of days required for sporogony
Recovery rate

because young children were assumed to dominate trans-
mission events [25]. However, clinically immune individ-
uals are infectious and can significantly affect the
dynamics of transmission when the distribution between
sensitive and resistant infections is different in nonim-
mune and clinically immune individuals.

Distributional differences in the frequencies of infection
types are driven by drug treatment rates, and these differ-
ences can be calculated as differences in the relative fitness
of resistant parasites compared with sensitive parasites.
Parasite fitness was calculated as the basic reproductive
number (R0) of both the resistant and the sensitive para-
sites when the population is completely naive (sub-
scripted by 1) and fully immune (subscripted by 2). The
R0-values of the sensitive parasites are

w bV(1-41) [ + c2 (7a)
Rowl ---0---- 61 + -------- (7a)
8+pl+*W+lw I P2+Tw+I

_R bVc2(1-2) (7b)
R,w2 P2+T (7b)

and the R0-values of the resistant strains are

Ro,x 1 + C (8a)
O+rx+P rx+P

RO,x2 = (8b)
rx+p f

Previous models suggested that the parasite with the high-
est R0 would be the most fit and should predominate
[27,40]. That conclusion always holds when rx= rw, c1 = c2,
and p, = p2. However, differences in the drug treatment
rates between clinically immune and nonimmune indi-
viduals can change the result, and it was found that resist-
ant parasites may not spread even when they have the

-1 10 years
Y- 2 years
,1 0.3
42 0.01
af, = P 0.025(0.2) = 1/200
2f2= P2 0.01(0.2) = 1/500
m 180/100000/year
a 0.3
b 0.8
g 1/10
n 10
rw 1/(165/b)
r. rw(fitness cost)

dominant Ro. This can be explained by calculating the
effective reproductive number of the resistant parasite, R,
where Rx = R,, .3 and d is the fraction of the population
that can be infected and will transmit the resistant infec-
tion. The effective reproductive number of the resistant
parasite is

Rx= (SS + S2)((1 -f) R,x +fR,x2)

where f is the fraction of the population that is clinically
immune, and S1 and S2 are the proportion of the popula-
tion that is susceptible and nonimmune, and susceptible
and clinically immune, respectively.

The results suggest that as the transmission rate increases,
the fraction of clinically immune individuals increases,
significantly reducing the effective reproductive number
of the resistant parasite. At certain parameter values it was
even possible for the effective reproductive number of the
resistant parasite to fall below 1 while the reproductive
number of the sensitive parasite remained above 1, effec-
tively abrogating the ability of resistant parasites to spread
(Figure 3). Thus, at higher transmission levels it may be
possible for the resistant parasite to have the dominant Ro
but be unable to spread because of the population-wide
level of immunity.

Though the ability of the resistant parasite to spread is not
always compromised, the generalizable result is that at all
levels of fitness cost (including none), it is easier for resist-
ant parasites to spread at lower levels of transmission (Fig-
ure 4). This reduction in the ability of individuals to
transmit the resistant parasite suggests an important
implication: it is relatively easier for resistance to spread in
an area of lower transmission than in an area of higher
transmission. This also helps explain, from an epidemio-
logical perspective, why resistance in general tends to
evolve faster in lower-transmission settings.

Page 5 of 9
(page number not for citation purposes)


r Co



3 -2 1 0 1 2 3 4
Log ofVectonal Capacity

Figure 2
Steady state infection level. The proportion of people
with P. falciparum infections increases extremely rapidly as
vectorial capacity increases, but there is a large shift between
those who are nonimmune and those who are immune.
Semi-immune individuals mirror the trajectory of total infec-
tions as vectorial capacity increases and quickly become the
large majority of infections. At low levels of vectorial capac-
ity, the proportion of the population that is clinically immune
remains extremely low. However, as vectorial capacity
increases, the proportion of individuals that are clinically
immune nonimmunee) increases (falls). The clinical incidence
rate increases sharply at first but then plateaus as the pro-
portion of the population that is clinically immune increases.
The clinical incidence rate is defined at equilibrium as the
sum of susceptible individuals who become infected with a
wild-type phenotype and develop an immediate clinical infec-
tion plus individuals already infected who become sympto-

matic h,S4, + li.s where i and j are the

phenotype of the infection and the immune stage, respec-

Simulations of the model found a significant relationship
between malaria immunity and the ability of resistant par-
asites to spread from an initial locus. Previous epidemio-
logical models of malaria transmission assumed that
immune individuals were not infectious; however, the
best available evidence in malaria suggests that immune
individuals remain infectious, though they transmit less
efficiently to mosquitoes [22-24,31] and the frequency
and severity of clinical disease declines [16-19]. Because

-3 -2 -1 0 1 2 3 4 5 6
Log of Vectorial Capacity

Figure 3
Immunity reduces resistant parasite fitness. The clini-
cally immune class influences the ability of a resistant parasite
to invade. Since individuals in this class do not progress to
clinical malaria as often, they are treated less often; this cre-
ates a refuge for the wild-type parasites. As vectorial capacity
increases, the clinically immune class is maintained at higher
and higher levels until it becomes biologically impossible for
the resistant parasite to spread. This paradigm exists because
of the population of clinically immune individuals, without
whom the resistant parasite would be able to spread at any
vectorial capacity (as shown by the dashed line above).

they remain infectious but asymptomatically infected,
their relatively lower usage of antimalarials creates a natu-
ral refuge for sensitive parasites, similar to the way non-
transgenic crops act as refugia for Bt-sensitive insects in
agriculture [41]. Increases in the transmission rate result
in concomitant increases in the proportion of clinically
immune individuals, which increases the size of the ref-
uge and reduces the ability of the resistant parasite to
spread. This suggests an epidemiological rationale for the
more likely emergence of resistance in low- or unstable-
transmission settings.

The presence of a refuge of clinically immune individuals
alters the ecological landscape that resistant parasites face,
and introduces a mechanism for the coexistence of resist-
ant and sensitive parasites. Simple models of directly
transmitted infections [40], and a previous model of
malaria transmission [27], suggested there was a critical
threshold level of treatment above which resistance would
fix and below which it would not spread. However, in
those models, there was either no immunity, or immune
individuals were assumed to be noninfectious and were,
therefore, irrelevant for transmission and selection for

Page 6 of 9
(page number not for citation purposes)

; Ro
resistant parasite Proportion clinically immune

,. re

------------------ ----------------------------------
-- -----
< resistant parasite

Malaria Journal 2008, 7:67

control project in the Garki region of Nigeria [35]. The
mathematical model of control developed for the Garki
Base project considered the clearance of infection in hosts that
have been infected by more than one parasite brood
20% [[20,45], and the development of blood-stage immunity
and transmission-blocking immunity in the human host
S33% [20]. Despite a model that assumed parasite infections
43% were complex and evidence that gametocytes remain
50% present throughout life, the Garki model made a simplify-
ing assumption that new infections remained infectious
for a short period of time. Thereafter, infections persisted
until the person either cleared the infection or developed
immunity. The Garki model called this a semi-immune
state. In this semi-immune state, humans could become
re-infected, but they were never infectious. Mathematical
epidemiologists later noted that individuals who were
.I. infected but not infectious were epidemiologically irrele-
-6 -4 -2 0 2 4 6 vant for transmission, and the Garki model was simplified
Vectorial Capacity into an SIRS compartment model in which the dynamics
4 of infection in the semi-immune population were
e reproductive rate and fitness cost of resist- replaced by a recovered and immune state with immune
*he higher the fitness cost of resistance (I rjr,), the boosting [25,26]. The SIRS model was a familiar model to
he absolute level of the effective reproductive mathematical epidemiologists, and it has been the basis
. Qualitatively, this suggests an important implication: for many subsequent papers on malaria epidemiology,
ays more difficult for the parasite to evolve (i.e., but the assumptions about immunity have been propa-
and spread) resistance in higher-transmission set- gated and revised without much critical thought.


resistance. Clinical immunity changes the criteria for
determining the critical threshold so that the fraction of
clinical episodes that are treated, as well as the level of
immunity in the community, which determines the frac-
tion of new infections that result in a clinical episode, are
both important. Therefore, in populations where immu-
nity has developed, the resistant parasite faces a different
fitness landscape; they may not be the most fit, despite
having the highest R0, which can prevent the parasite from
either fixing or spreading. This result also suggests that dif-
ferential treatment levels of host-groups can create differ-
ential ecological niches allowing the resistant and
sensitive parasites to coexist [42].

History of epidemiological models of malaria
There is a striking difference between the findings of this
paper and a previously published epidemiological model
of antimalarial resistance [27]. To explain the differences,
a brief history of malaria models is warranted. A quantita-
tive approach to malaria epidemiology organized around
the parasite life-cycle was first described mathematically
by Ross [43], and later revised by Macdonald to consider
mosquito mortality during sporogony [32,44]. In the
1970s several major innovations in malaria modeling
were first introduced and field-tested during a malaria

This paper is an attempt to demonstrate the importance of
immunity in modeling the transmission dynamics of
malaria with respect to the introduction of resistance.
While the implications of the results are significant for
drug policy, it is prudent to note that as with all models a
number of simplifying assumptions have been made. In
incorporating immunity, the process of immunity acqui-
sition has been simplified so that individuals acquire
immunity in a stepwise fashion after 10 years of continu-
ous infection, and they lose immunity after two years of
continuous non-infection. This is a simplification of the
complicated process that is immunity. It ignores age
effects, differences between immune individuals at older
ages and younger ages, and faster acquisition of immu-
nity, which may occur in areas of higher transmission [46]
or after fewer challenges in low transmission areas [47].
While this is a simplification, evidence suggests that it
gives a representation of reality that is close enough to
make qualitative observations [29,30,34,35]. Despite the
simplification of immunity, the basic model (without
resistance) accords well with both field data and other
models of transmission dynamics, and changes in these
values did not change the qualitative results. Thus, the
model, as formulated, likely captures the qualitative dif-
ferences that exist between areas with different transmis-
sion levels, and is, hopefully, only the first step in more

Page 7 of 9
(page number not for citation purposes)




ance. T
lower ti
it is alw

Malaria Journal 2008, 7:67

detailed modeling of the spread of antimalarial resistance
which takes account of the importance of immunity.

The link between the transmission rate and the probabil-
ity that de novo mutants can both arise and spread is a crit-
ical issue for developing strategies to mitigate resistance
emergence. Though it is impossible to be certain, histori-
cal analysis of the evolution of resistance has suggested
that it originated in low or unstable transmission areas
[2,3]. Quantitative population genetic models have sup-
ported this assertion, suggesting that there may be genetic
reasons why resistance is less likely to evolve in high-
transmission settings [3,9,11,12], but individual models
may not always explain population-level phenomena.
Epidemiological models, which can take account of spe-
cific population characteristics, have been hampered by
the assumption that immune individuals play no qualita-
tive role in the dynamics of transmission. Consequently,
an epidemiological model could not offer an explanation
for why resistance emerged outside Africa [27]. The model
in this paper demonstrates that the existence of a refuge
for drug-sensitive parasites in high-transmission areas can
slow or prevent the evolution of resistance; this has enor-
mous implications for the design of effective malaria con-
trol strategies.

Efforts to control malaria have been undermined by the
emergence of resistance, which now exists to all known
antimalarials except the artemisinin compounds. To help
ensure a longer period of efficacy for artemisinin, the
World Health Organization has proposed a global sub-
sidy to comply with its mandate that all new artemisinin-
based therapies be deployed as combinations [48]. One
concern with this strategy is that increased use in high-
transmission areas may engender resistance at a faster rate.
The results as determined by this model, mitigates that
concern, showing that the existence of a refuge of immune
individuals makes this outcome highly unlikely. How-
ever, in areas with low transmission intensity, the relative
paucity of immune individuals increases the risk that
resistance will emerge. Because the emergence of resist-
ance to an antimalarial in any country threatens the via-
bility of the drug in all countries [49], areas of relatively
lower transmission should be a focus in controlling the
emergence of resistance. As a corollary, eliminating
malaria from areas of low transmission intensity may
have the global benefit of prolonging the effective lifetime
of antimalarial drugs.

Although areas of low transmission should implement
strategies to reduce the likelihood of resistance, high-
transmission areas are also a concern because transmis-
sion intensity is reduced through mass distribution of
insecticide-treated bed nets. The analysis in this paper sug-

gests that among-host selection for resistance increases as
functional immunity wanes and the refuge for drug-sensi-
tive parasites shrinks or disappears. Thus, implementation
of transmission reduction strategies should also include
improved surveillance for drug resistance with resources
to deploy appropriate containment strategies to prevent
the geographical dissemination of resistance.

The intention of this study was to evaluate the qualitative
importance of immunity in the transmission dynamics of
malaria and its role in the development of resistance. Epi-
demiological models of malaria transmission, which can
synthesize various factors about how emergence and
spread are linked to the composition of the population,
can be useful tools in determining optimal drug treatment
strategies to extend the effective life of newly introduced
therapies; however, they must take account of immune
individuals and the refuge they provide for sensitive para-

Authors' contributions
DLS and EYK developed the model. All authors contrib-
uted to the analysis and writing and have read and
approved the final version of the manuscript.

We are grateful to Peter Billingsley and Sunetra Gupta for their input on
the design of the model as well as Karen I Barnes, Anders Bjorkman, lan C
Boulton, Karen P Day, Abdoulaye Djimde, Arjen M Dondorp, Ogobara
Doumbo, Rose McGready, Malcolm Molyneux, Myaing M Nyunt, Christo-
pher V Plowe, Ric N Price, Allan Schapira and Nicholas J White for their
comments on the model at the Gordon Research Conference on Malaria.
This research was supported by a grantfrom the Bill & Melinda Gates Foun-
dation. Decisions concerning the study design; the collection, analysis, and
interpretation of data; the writing of the manuscript; and the decision to
submit the manuscript for publication, were made entirely by the authors
with no input from the Bill & Melinda Gates Foundation.

I. Mathers CD, Lopez AD, Murray CJL: The burden of disease and
mortality by condition: data, methods, and results for 2001.
In Global Burden of Disease and Risk Factors Edited by: Lopez AD, Math-
ers CD, Ezzati M, Jamison DT, Murray CJL. Washington, DC The
International Bankfor Reconstruction and Development/ The World
Bank; 2006.
2. White NJ: Preventing antimalarial drug resistance through
combinations. Drug Resist Updat 1998, 1:3-9.
3. White NJ, Pongtavornpinyo W: The de novo selection of drug-
resistant malaria parasites. Proc R Soc Lond B 2003,
4. Wellems TE, Plowe CV: Chloroquine-resistant malaria. j Infect
Dis 2001, 184(6):770-776.
5. Wootton JC, Feng X, Ferdig MT, Cooper RA, Mu J, Baruch DI, Magill
AJ, Su XZ: Genetic diversity and chloroquine selective sweeps
in Plasmodium falciparum. Nature 2002, 4 18(6895):320-323.
6. Cortese JF, Caraballo A, Contreras CE, Plowe CV: Origin and dis-
semination of Plasmodium falciparum drug-resistance
mutations in South America, J Infect Dis 2002, 186(7):999-1006.
7. Nair S, Williams JT, Brockman A, Paiphun L, Mayxay M, Newton PN,
Guthmann JP, Smithuis FM, Hien TT, White NJ: A selective sweep
driven by pyrimethamine treatment in southeast Asian
malaria parasites. In Mol Biol Evol Volume 20. Issue 9 SMBE;

Page 8 of 9
(page number not for citation purposes)

Malaria Joumnal 2008, 7:67

8. Roper C, Pearce R, Bredenkamp B, Gumede J, Drakeley C, Mosha F,
Chandramohan D, Sharp B: Antifolate antimalarial resistance in
southeast Africa: a population-based analysis. In Lancet Volume
361. Issue 9364 Elsevier; 2003:1174-1181.
9. Hastings IM: A model for the origins and spread of drug-resist-
ant malaria. Parasitology 1997, I 15:133-141.
10. Plowe CV, Kublin JG, Doumbo OK: P. falciparum dihydrofolate
reductase and dihydropteroate synthase mutations: epide-
miology and role in clinical resistance to antifolates. Drug
Resist Updat 1998, 1:389-396.
II. Dye C, Williams B: Multigenic drug resistance among inbred
malaria parasites. In Proc R Soc Lond B Volume 264. Issue 1378 The
Royal Society; 1997:61-67.
12. Paul RE, Packer MJ, Walmsley M, Lagog M, Ranford-Cartwright LC,
Paru R, Day KP: Mating patterns in malaria parasite popula-
tions of Papua New Guinea. Science 1995, 269(523 I): 1709.
13. Gatton ML, Hogarth W, Saul A: Time of treatment influences
the appearance of drug-resistant parasites in Plasmodium
falciparum infections. In Parasitology Volume 123. Issue 06 Cam-
bridge Univ Press; 2003:537-546.
14. Hay SI, Guerra CA, Tatem AJ, Atkinson PM, Snow RW: Urbaniza-
tion, malaria transmission and disease burden in Africa. Nat
Rev Microbiol 2005, 3(1):81-90.
15. Smith DL, McKenzie FE, Snow RW, Hay SI: Revisiting the basic
reproductive number for malaria and its implications for
malaria control. PloS Biol 2007, 5(3):e42.
16. Marsh K, Snow RW: Malaria transmission and morbidity. Par-
assitologia 1999, 41(1-3):241-246.
17. Smith T, Killeen G, Lengeler C, Tanner M: Relationships between
the outcome of Plasmodium falciparum infection and the
intensity of transmission in Africa. Am J Trop Med Hyg 2004,
71((Suppl 2)):80-86.
18. Baird JK: Host Age as a determinant of naturally acquired
immunity to Plasmodium falciparum. Parasitology Today 1995,
11(3):105-1 II.
19. McGregor IA, Smith DA: A health, nutrition and parasitological
survey in a rural village in West Kiang. Trans R Soc Trop Med
Hyg 1952, 46(4):403-427.
20. Dietz K, Molineaux L, Thomas A: A malaria model tested in the
African savannah. Bull World Health Organ 1974, 50(3-4):347-357.
21. Carter R, Bushell G, Saul A, Graves PM, Kidson C: Two apparently
nonrepeated epitopes on gametes of Plasmodium falci-
parum are targets of transmission-blocking antibodies. Infect
Immun 1985, 50(1):102-106.
22. Bonnet S, Gouagna LC, Paul RE, Safeukui I, Meunier JY, Boudin C:
Estimation of malaria transmission from humans to mosqui-
toes in two neighboring villages in south Cameroon: evalu-
ation and comparison of several indices. Trans R Soc Trop Med
Hyg 2003, 97(1):53-59.
23. Githeko AK, Brandling-Bennett AD, Beier M, Atieli F, Owaga M, Col-
lins FH: The reservoir of Plasmodium falciparum malaria in a
holoendemic area of western Kenya. Trans R Soc Trop Med Hyg
1992, 86(4):355-358.
24. Trape JF, Rogier C, Konate L, Diagne N, Bouganali H, Canque B,
Legros F, Badji A, Ndiaye G, Ndiaye P: The Dielmo Project: a lon-
gitudinal study of natural malaria infection and the mecha-
nisms of protective immunity in a community living in a
holoendemic area of Senegal. Am J Trop Med Hyg 1994,
25. Aron JL: Acquired immunity dependent upon exposure in an
SIRS epidemic model. Math Biosc 1988, 88(1):37-47.
26. Aron JL: Mathematical modeling of immunity to malaria.
Math Biosc 1988, 90:385-396.
27. Koella JC, Antia R: Epidemiological models for the spread of
anti-malarial resistance. Malar 2003, 2:3.
28. Baird JK, Jones TR, Danudirgo EW, Annis BA, Bangs MJ, Hasan Basri
P, Masbar S: Age-dependent acquired protection against Plas-
modium falciparum in people having two years exposure to
hyperendemic malaria. Am J Trop Med Hyg 1991, 45(165-76
29. Reyburn H, Mbatia R, Drakeley C, Bruce J, Carneiro I, Olomi R, Cox
J, NkyaWMMM, Lemnge M, Greenwood BM, Riley EM: Association
of transmission intensity and age with clinical manifestations
and case fatality of severe Plasmodium falciparum malaria.
JAMA 2005, 293(12):1461-1470.

30. Snow RW, Omumbo JA, Lowe B, Molyneux CS, Obiero JO, Palmer
A, Weber MW, Pinder M, Nahlen B, Obonyo C: Relation between
severe malaria morbidity in children and level of Plasmo-
dium falciparum transmission in Africa. Lancet 1997,
31. Dal-Bianco MP, Koster KB, Kombila UD, Kun JFJ, Grobusch MP,
Ngoma GM, Matsiegui PB, Supan C, Salazar CLO, Missinou MA, Issi-
fou S, Lell B, Kremsner P: High prevalence of asymptomatic
Plasmodium falciparum infection in Gabonese adults. Am J
Trop Med Hyg 2007, 77(5939-942 [
32. Macdonald G: The Epidemiology and Control of Malaria. Lon-
don, United Kingdom Oxford University Press; 1957.
33. Smith DL, McKenzie FE: Statics and dynamics of malaria infec-
tion in Anopheles mosquitoes. Malar] 2004, 3:13.
34. Smith DL, Guerra CA, Snow RW, Hay SI: Standardizing estimates
of the Plasmodium falciparum parasite rate. Malar J 2007,
35. Molineaux L, Gramiccia G: The Garki Project. Research on the
Epidemiology and Control of Malaria in the Sudan Savanna
of West Africa. Geneva, World Health Organization; 1980.
36. Hayward R, Saliba KJ, Kirk K: pfmdrl mutations associated with
chloroquine resistance incur a fitness cost in Plasmodium fal-
ciparum. Molecular Microbiology 2005, 55(4):1285-1295.
37. Hastings IM, Donnelly MJ: The impact of antimalarial drug
resistance mutations on parasite fitness, and its implications
for the evolution of resistance. Drug Resist Updat 2005, 8(1-
38. Laufer MK, Plowe CV: Withdrawing antimalarial drugs: impact
on parasite resistance and implications for malaria treat-
ment policies. In Drug Resist Updat Volume 7. Issue 4-5 Elsevier;
39. Djimde AA, Doumbo OK, Traore O, Guindo AB, Kayentao K,
Diourte Y, Niare-Doumbo S, Coulibaly D, Kone AK, Cissoko Y:
Clearance of drug-resistant parasites as a model for protec-
tive immunity in Plasmodium falciparum malaria. Am] Trop
Med Hyg 2003, 69(5558 [
14695097]. ASTMH
40. Bonhoeffer S, Lipsitch M, Levin BR: Evaluating treatment proto-
cols to prevent antibiotic resistance. PNAS 1997,
94(22):12106-121 II.
41. Laxminarayan R: Battling resistance to antibiotics and pesti-
cides. Washington, DC Resources for the Future; 2003.
42. Levin SA: Community equilibria and stability, and an exten-
sion of the competitive exclusion principle. In Am Nat Volume
104. Issue 939 JSTOR; 1970:413-423.
43. Ross R: The Prevention of Malaria, 2nd ed. London Murray;
19 11.
44. Macdonald G: The analysis of the sporozoite rate. Trop Dis Bull
1952, 49(6):569-586.
45. Bailey NTJ: The Biomathematics of Malaria. London, C. Griffin
& Co.; 1982.
46. Snow RW, Nahlen B, Palmer A, Donnelly CA, Gupta S, Marsh K: Risk
of severe malaria among African infants: direct evidence of
clinical protection during early infancy. j Infect Dis 1998,
47. Baird JK, Krisin, Barcus MJ, Elyazar IR, Bangs MJ, Maguire JD, Fryauff
DJ, Richie TL, Sekartuti, Kalalo W: Onset of clinical immunity to
Plasmodium falciparum among Javanese migrants to Indo-
nesian Papua. Ann Trop Med Parasitol 2003, 97(6):557-564.
48. World Health Organization: Antimalarial drug combination
therapy: report of a technical consultation. Geneva World
Health Organization; 2001.
49. Klein E, Laxminarayan R, Smith DL, Gilligan C: Economic incen-
tives and mathematical models of disease. Environ Dev Econ
2007, 12(5):707-732.

Page 9 of 9
(page number not for citation purposes)

Malaria Journal 2008, 7:67

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs