A recent paper by Gandon et al. presents a model for how leaky
vaccines can lead to evolutionary changes in the virulence of parasites.
Whilst readers of some of the press reaction to this paper might be
forgiven for thinking that it referred to certainties about actual
malaria vaccines, there are currently no malaria vaccines beyond the
early clinical trial phase, so the paper could in principle offer no
more than theoretical predictions. The models might well be less
appropriate for the chosen example of Plasmodium falciparum
malaria than for other parasites. Critical review of the Gandon et
al. paper highlights both the fact that the determinants of
virulence in malaria are not well understood, and that monitoring of
malaria vaccines should consider more than just the immediate effects on
vaccine efficacy.
There is an important place in science for conjectures, and the example
of malaria vaccines is justifiably topical. Plasmodium falciparum
is the most important eukaryotic parasite of humans, with hundreds of
millions of people infected at any one time, and at least 1 million
people dying as a consequence each year. The need for a malaria vaccine
is manifest, but none are currently registered for use, and the
complexity of the strategies used by the parasite to evade the host
immune responses makes for major technical difficulties in designing one
that is effective. It is likely that we will need to make do with an
imperfect vaccine that allows some malaria parasites to survive, even in
vaccinated hosts.
Much vaccination theory assumes the objective to be pathogen
eradication. With malaria, this could well be an unrealistic goal, but
the disease is so frequent that even a partially effective vaccine
delivered via the less-than-optimal health systems available in
sub-Saharan Africa could have a major impact on public health. To the
epidemiologist, the benefits of such a vaccine would be obvious.
However, the evolutionary consequences of selection induced by such
vaccines are not at all obvious.
A new paper by Gandon et al.
[1] represents an attempt to predict some of the
consequences of such vaccines. The authors use a mathematical approach
combining simple models of parasite transmission with a model for the
evolution of virulence. The basis of this model is cost-of-virulence
theory [2], one of several hypotheses for why
host–pathogen interactions do not evolve towards symbiosis
[3]. The theory acknowledges that killing the host
is bad for the parasite, but that if more-virulent parasites have a
higher transmission rate, competition among parasites for hosts will
lead to tradeoffs with a stable state at an intermediate level of
virulence.
Most effort in malaria vaccine research so far has been to develop
vaccines to prevent infection, one of the kinds of vaccine considered by
Gandon et al. An example is the RTS,S construct, which recently
demonstrated some efficacy in The Gambia [4]. Gandon
et al. suggest that the use of such a vaccine will lead to
selection for reduced virulence. One way in which this could happen is
if the virulent parasite has a high intrinsic growth rate. Competition
among parasites leads to selection of more virulent parasites, because
the faster growing parasites corner the resources of the host for
themselves, pulling the evolutionarily stable state further towards
virulence than in the situation without competition. A vaccine that
reduces transmission will reduce competition among different malaria
parasites and hence select for less virulent parasites.
However, natural immunity against P. falciparum does not prevent
infection, and so it is unclear whether it is possible to completely
block infection with a vaccine. Other vaccination strategies that mimic
more closely natural immunity are therefore also being tried. These
include vaccines that limit the multiplication of the blood stages of
the parasite, thus preventing the parasites reaching densities at which
they cause severe illness or death.
One evolutionary consequence of such a vaccine against a polymorphic
parasite is selection in favour of pathogens expressing alleles
different from those in the vaccine. There is ample evidence for such
vaccine-induced selection from pathogens other than malaria and evidence
for such selection has already emerged in one trial of a malaria vaccine
[5]. Therefore, the vaccine must be designed to
contain epitopes that cover a broad range of parasite types, including
those causing severe disease. (Extending this idea, we can speculate
that, if the specific factors affecting parasite virulence were
understood, one might be able to induce immune responses that directly
select in favour of less virulent parasites.)
However, Gandon et al. do not discuss this kind of selection, but
rather the selection that occurs if the target epitope of the vaccine is
not itself responsible for virulence. Within the cost-of-virulence
model, the level of virulence results from a tradeoff between virulence
and persistence, so the effect of any intervention that makes the
parasite less virulent in practice is to generate selection in favour of
intrinsic virulence until the equilibrium is regained. Any vaccine that
slows parasite growth will lead to evolution in the direction of greater
intrinsic virulence, so that unvaccinated individuals will be at greater
risk than before.
The evidence for this cost-of-virulence model is limited. The theory
assumes that the rate of parasite transmission from the human host
increases with parasite virulence. This makes sense if virulence and
transmission are simple functions of parasite multiplication rates. The
Edinburgh group, where Gandon et al. are based, has shown that
faster growing parasites are more virulent in the P. chabaudi–rodent
model system. Chotivanich et al. [6] have
elegantly demonstrated a relationship between the in vitro growth
rate of parasites and the severity of disease in P. falciparum
patients in Thailand (a country with relatively low malaria endemicity).
However, this is only one group of patients, who are likely to be very
different from the children in endemic areas of Africa who are the main
victims of the disease. Even in the Thai study, there was considerable
overlap in the growth rates of parasites in uncomplicated malaria with
those of severe cases. Virulence in malaria is certainly not merely a
matter of intrinsic growth rate, even in previously unexposed subjects
[7].
In areas of endemic malaria in sub-Saharan Africa, in contrast to the
infections in Thai patients, most P. falciparum infect people
without causing many overt symptoms. The overall disease burden is
enormous, but most of it occurs in specific sections of the population
(young children, pregnant women, and short-term visitors). In the
remainder, the parasites persist at low densities, but do not often
cause acute illness.
In such a situation, competition among malaria parasites must be
mediated by immunological mechanisms. Without immunological modulation,
the progeny of a single inoculation would proliferate in an uncontrolled
manner, killing the host within a few weeks. Persistence and
transmission are mainly functions of how the parasite deals with this
immune modulation, and the intrinsic growth rate is only one, probably
rather minor, factor in this. But, although we have some idea about the
strategies of immunoevasion (extreme polymorphism, antigenic switching,
smokescreen effects etc.), we have very little idea about how these come
together to define whether a given parasite persists in a given host,
and even less how they affect transmission or virulence. There are many
possible speculations about how immunity might affect the evolutionary
biology of P. falciparum. For example, an attractive strategy
from the point of view of a parasite already established within a host,
and of the host itself, would be for the parasite to co-opt the host
immune system to fight off potential competitors. There is some evidence
that this happens. In particular, there is an upper limit to the
incidence of malaria infections as the inoculation rate increases, and
there is some evidence that the pre-existing infections might protect
against superinfections in partially immune hosts [8].
What would be the implications of vaccination against blood-stage
antigens if they play a role in cross protecting against invading
clones? It is not easy even to predict the immediate consequences, which
might well depend upon whether vaccination tends to eliminate the
existing parasites, or to boost immune responses to the existing
infections. The evolutionary implications are at least as unclear (
Box 1).
It is commendable that Gandon et al. have stimulated thinking
about the possible evolutionary consequences of malaria vaccines. But it
is important to remember that their model is only one of many that are
possible and that it might be a better model for simpler systems (such
as viral pathogens) than it is for malaria. Unexpected effects of
vaccination are probable when imperfect malaria vaccines are introduced
on a large scale. The secondary effects of vaccination should certainly
be monitored closely and, as Gandon et al. recommend, intrinsic
growth rates of parasites are one parameter that should not be
forgotten. But, it is not at all obvious that the predictions that they
make will prove accurate.
