Virulence evolution in response to vaccination: the case of malaria (Invited paper for the journal Vaccine summarising a workshop at Rutgers University in July 2005 on Vaccination and its evolutionary consequences.)

One theory of why some pathogens are virulent (i.e. they damage their host) is that they need to extract resources from their host in order to compete for transmission to new hosts, and resource extraction can damage the host. Here we describe our studies in malaria that test this idea. We go on to show that host immunity can exacerbate selection for virulence and therefore that vaccines that reduce pathogen replication may select for more virulent pathogens, eroding the benefits of vaccination and putting the unvaccinated at greater risk. We suggest that in disease contexts where wild-type parasites can be transmitted through vaccinated hosts, evolutionary outcomes need to be considered. 1. An evolutionary hypothesis for pathogen virulence Why are pathogens virulent 1 ? Why would they run the risk of killing their host when, in doing so, they lose their ongoing source of transmission to new hosts? Some evolutionary biologists believe that the answer to this question will make it possible to design vaccines and other control measures that, in the event of eradication being impossible, drive the pathogen towards lower virulence (Williams and Nesse, 1991; Dieckmann et al., 2002). One answer to this question is that virulence is a mistake by the pathogen – an ultimately maladaptative outcome that occasionally happens when a pathogen accidentally ends up in an abnormal host environment, or when a virulent mutant has a transient competitive advantage within a host (‘short-sighted, or dead-end evolution’) (Levin and Bull, 1994). An alternative (but not mutually exclusive (Frank, 1996)) idea is that the level of pathogen virulence observed in nature is a welladapted outcome of both positive and negative selective forces acting on virulence. Under this hypothesis, it is reckoned that to balance the fitness cost to the pathogen of host death, there must also be a virulence-related advantage to its fitness. This is the so-called ‘virulence trade-off hypothesis’. 1 Throughout this paper, we strictly define virulence as the fitness cost that the parasite causes the host. This may be through mortality, or morbidity-related reduction in fertility or fecundity. We sometimes use morbidity as a surrogate measure of virulence. Mackinnon, M.J., Gandon, S., & Read, A.F. Virulence evolution in response to vaccination: the case of malaria. Vaccine (in press) 2 Of all the explanations for virulence, the trade-off hypothesis has received most attention and a large body of theory has been derived from it. Yet it is poorly supported by data. Here we describe our studies in malaria parasites, the causative agents of a disease of global importance, in which we have comprehensively explored the trade-off hypothesis. We begin by summarising our experimental tests in a laboratory mouse-malaria system of the assumptions underlying the trade-off theory. We then ask whether the rodent data are relevant to malaria parasites in their human setting. Next, we use the trade-off theory to predict what the impact might be on the evolution of the pathogen’s virulence if malaria vaccines went into widespread use. Finally, we summarise an experimental evolution study to test our prediction that enhanced immunity would select for more virulent parasites. Together, this work has led us to a deeper understanding of why malaria still kills its host despite millenia of coevolution, and what might happen when disease control campaigns change the level of population immunity, e.g., enhance it using vaccines, or reduce it using bednets and vector control. 1.1. The trade-off hypothesis and its assumptions Under the trade-off hypothesis, it is assumed that there are both fitness benefits and costs associated with virulence. The cost is assumed to be host death because, for most pathogens, dead hosts do not transmit. The benefits associated with virulence are assumed to be production of more transmission forms per unit time, and/or increased persistence in a live host. However, the benefits of higher transmissibility and persistence only accrue if the host survives: if it dies, transmission immediately ceases, thereby directly reducing the pathogen’s fitness. The pathogen is thus playing a perilous game, attempting to maximise transmissibility and infection length while also keeping its host alive. Pathogens with the highest fitness are those with an intermediate level of virulence which balances these opposing contributions to fitness (Fig. 1). 1.2. Evidence from myxomatosis Although the trade-off idea was mooted by medical epidemiologists early in the 20 th century (Topley, 1919) and much later by theoretical biologists (Levin and Pimentel, 1981; Bremermann and Pickering, 1983), it was Anderson and May’s (1982) analysis of the principle using a real-life example which really focussed thinking. The myxoma virus was extremely virulent when released into rabbit populations in Australia and Britain in the early 1950’s, but over the next decade, the virus evolved to an intermediate and stable level of virulence (Fenner, 1953; Fenner et al., 1957). Using data from Fenner and colleagues (Fenner and Ratcliffe, 1965) which showed that viral strains that caused more host death were also those capable of generating longer infections (Fig. 2a), Anderson and May (1982) showed that maximum fitness of the virus occurred at an intermediate level of virulence. This was because the more virulent strains produced longer infections in the absence of host death, but death occurred more often in these strains. In their analysis they assumed that transmissibility was constant, i.e., the trade-off was generated exclusively by the relationship between virulence and infection length. Similarly, using the data of Mead-Briggs and Vaughan (1975) which show higher transmissibility in the more virulent strains (except at very high levels, Fig. 2b), Massad (1987) showed that maximum pathogen fitness was reached at intermediate levels of virulence. Thus Mackinnon, M.J., Gandon, S., & Read, A.F. Virulence evolution in response to vaccination: the case of malaria. Vaccine (in press) 3 either the virulence-infection length relationship or the virulence-transmissibility relationship, or the combination of the two (Fig. 2c) could explain the observed evolution of the myxoma virus from extreme virulence to intermediate virulence through time.