The three Ts of virulence evolution during zoonotic emergence

Abstract

There is increasing interest in the role that evolution may play in current and future pandemics, but there is often also considerable confusion about the actual evolutionary predictions. This may be, in part, due to a historical separation of evolutionary and medical fields, but there is a large, somewhat nuanced body of evidence-supported theory on the evolution of infectious disease. In this review, we synthesize this evolutionary theory in order to provide a framework for clearer understanding of the key principles. Specifically, we discuss the selection acting on zoonotic pathogens' transmission rates and virulence at spillover and during emergence. We explain how the direction and strength of selection during epidemics of emerging zoonotic disease can be understood by a three Ts framework: trade-offs, transmission, and time scales. Virulence and transmission rate may trade-off, but transmission rate is likely to be favoured by selection early in emergence, particularly if maladapted zoonotic pathogens have 'no-cost' transmission rate improving mutations available to them. Additionally, the optimal virulence and transmission rates can shift with the time scale of the epidemic. Predicting pathogen evolution, therefore, depends on understanding both the trade-offs of transmission-improving mutations and the time scales of selection.

Keywords: emerging zoonotic disease; evolution; trade-offs; transmission; virulence.

Figure 1.

The three Ts of virulence evolution during zoonotic emergence. Trade-offs between virulence and transmission rate determine pathogen fitness at every point during an epidemic, regulating pathogen fitness at the spillover barrier and shaping selection as the epidemic progresses. Early in the epidemic, however, individual transmission rate improving mutations may be ‘costless’ and not have trade-offs. Improvements in transmission rate are the most important selection pressure during epidemic take-off and building phases, though selection is weak at take-off. Finally, the time scale of the epidemic shifts the pathogen's optimal virulence and transmission rate strategies as the density of susceptible hosts changes. Created with Biorender.com. (Online version in colour.)

Figure 2.

Disease triangle of virulence. (Online version in colour.)

Figure 3.

Conceptual diagram of the Pareto front between virulence and transmission rate. A Pareto front between virulence and transmission rate defines a region of accessible phenotype space. Theory determines where the ‘optimal strategy’ sits on the Pareto front to determine which regions of this phenotype space are selectively advantaged or disadvantaged. Phenotype combinations far from the Pareto front may technically be possible but would be highly selectively disadvantaged and likely to go extinct. Possible phenotypes can move towards their optimal strategy along any pathway within the accessible phenotype space. However, we cannot know where a hypothetical phenotype sits below its individual Pareto front. Selection for improved transmission rate can, therefore, involve decreases, no changes, or increases in virulence depending on the pathogen's starting point and mutational availability. (Online version in colour.)

Figure 4.

Recently emerged viral zoonoses loosely follow a Pareto front of virulence and R₀ where R₀ seems to be maximized at intermediate case fatality rates (CFRs) within viral families. Data are from a dataset published in 2019 of recently emerged viral zoonoses from mammalian hosts [40]. Approximate R₀ is classified from 1 (no human-to-human transmission) to 4 (endemic transmission). In figure 4a, dots represent plotted residuals from linear models of CFR and approximate R₀ including virus family as a factor. By regressing out virus family, we somewhat control for the variation in trade-off shape for each virus and can make general observations across the dataset. Each dot, therefore, represents the virulence and R₀ of an individual epidemic of viral zoonosis scaled by virus family. In figure 4b, CFR and approximate R₀ are directly plotted and separated by virus family so that the non-aggregated trends could be seen within virus families. In both panels, dots are coloured by the phylogenetic distance between humans and the reservoir host. Plots were made with ‘ggplot2’. See electronic supplementary material for code. (Online version in colour.)