Epidemic spreading under mutually independent intra- and inter-host pathogen evolution

Xiyun Zhang, Zhongyuan Ruan, Muhua Zheng, Jie Zhou, Stefano Boccaletti, Baruch BarzelarXiv 2021 · DOI · arXiv

TL;DR

This paper introduces a new modeling framework that couples network-based epidemic spreading with the stochastic evolutionary dynamics occurring within individual hosts. It reveals a "mutation-driven phase" where pathogens that are initially too weak to cause a pandemic can evolve and "break through" into widespread prevalence, provided their mutation rate falls within a specific "Goldilocks" zone.

What Problem Does This Paper Address?

In classical epidemiology, the fate of an outbreak is usually decided by a single number: $R_0$ (the basic reproduction number). If $R_0 > 1$, the disease spreads; if $R_0 < 1$, it dies out. However, this assumes the pathogen is a static entity. In reality, viruses like SARS-CoV-2 or influenza evolve as they spread.

The challenge is that intra-host dynamics (mutation and selection inside a single body) happen at a different scale and follow different rules than inter-host dynamics (spreading through a social network). Most models treat these separately. The authors address this gap by creating a unified framework to predict when a sub-pandemic pathogen will undergo a "breakthrough mutation" and how the timing of our public health response affects this evolutionary risk.

Methods

The researchers developed a modeling framework that combines two distinct processes:

  1. Inter-host Spreading: They used a standard Susceptible-Infected-Recovered (SIR) model on an Erdős-Rényi random network.
  2. Intra-host Evolution: They modeled the pathogen's internal replication as a stochastic process. Crucially, they simplified the complex genetic landscape into an effective random walk in "fitness space."

The model tracks two types of fitness: intra-host fitness (how well the virus replicates in a body) and inter-host fitness (how easily it jumps to a neighbor). By assuming these two are often independent, they could characterize the evolutionary pressure using a single parameter, $\sigma$ (the observed mutation rate). They then derived analytical equations to define the boundaries of different pandemic phases based on the balance between the timescale of epidemic decay and the timescale of evolutionary breakthrough.

Datasets

As a theoretical physics and complexity science paper, the primary "data" comes from extensive numerical simulations on synthetic networks ($N=5,000$ to $N=10^8$ nodes).

However, to ground their work in reality, they used empirical data from the SARS-CoV-2 infection cycle (incubation periods, symptomatic vs. asymptomatic ratios) and observed spreading rates from 12 different countries to calibrate their reinfection and immunity-evasion models.

Key Insights and Ideas

  • The Goldilocks Zone: Pathogens need a specific range of mutation rates to thrive. If mutations are too slow, the virus dies out before it can improve. If they are too fast, the pathogen enters a "Volatile Phase" where it mutates so erratically that it can't "lock in" advantageous traits, leading to extinction.
  • Explosive Transitions: Unlike standard epidemic models where an outbreak grows gradually as $R_0$ increases, mutation-driven outbreaks are first-order (discontinuous) transitions. This means a system can appear stable and then suddenly "explode" into a pandemic with almost no warning.
  • Time Sensitivity and the "Pool of Opportunity": The risk of a breakthrough mutation is directly tied to the number of infected individuals ($I_0$). A larger pool of infected people provides more "tickets in the evolutionary lottery."
  • Late Mitigation is Doubly Dangerous: If we wait too long to implement social distancing, we aren't just dealing with more cases; we are actually shifting the phase boundary. A late response allows the virus to reach a high prevalence where a breakthrough mutation becomes almost inevitable, potentially leading to a second wave that is harder to control than the first.

Key Results

  • Phase Diagram: The authors identified four distinct regimes in the $(\sigma, R_0)$ plane: Infection-free, Mutation-driven, Volatile, and Pandemic.
  • Critical Mutation Rate ($\sigma_c$): They derived a scaling law (Eq. 15) that predicts the minimum mutation rate needed for a breakthrough. It shows that $\sigma_c$ is inversely proportional to the logarithm of the initial infected population ($\ln(I_0)$).
  • Hysteresis: They demonstrated that the path to eradicating a virus is different from the path it took to emerge. Once a virus has evolved into the mutation-driven phase, simply pushing $R_0$ back to its original sub-critical level may not be enough to stop it.

Limitations and Future Work

The authors acknowledge several simplifications:

  • Fitness Independence: The baseline model assumes intra-host and inter-host fitness are uncorrelated. While they explore correlated fitness in the supplements, real-world trade-offs (e.g., a virus that replicates faster might make the host too sick to travel) could be more complex.
  • Network Topology: The study primarily uses random networks. Real-world social networks have "small-world" properties and heavy-tailed degree distributions that might alter the speed of the transitions.
  • Homogeneity: The model assumes a uniform mutation rate $\sigma$ across all hosts, ignoring "superspreaders" or individuals with compromised immune systems who might act as evolutionary "reactors" for the virus.

Future work could involve applying this framework to specific drug-based mitigation strategies, where the "arms race" between treatment and viral resistance creates a biased (rather than random) walk in fitness space.

Summary updated

Abstract

The dynamics of epidemic spreading is often reduced to the single control parameter $R_0$, whose value, above or below unity, determines the state of the contagion. If, however, the pathogen evolves as it spreads, $R_0$ may change over time, potentially leading to a mutation-driven spread, in which an initially sub-pandemic pathogen undergoes a breakthrough mutation. To predict the boundaries of this pandemic phase, we introduce here a modeling framework to couple the network spreading patterns with the intra-host evolutionary dynamics. For many pathogens these two processes, intra- and inter-host, are driven by different selection forces. And yet here we show that even in the extreme case when these two forces are mutually independent, mutations can still fundamentally alter the pandemic phase-diagram, whose transitions are now shaped, not just by $R_0$, but also by the balance between the epidemic and the evolutionary timescales. If mutations are too slow, the pathogen prevalence decays prior to the appearance of a critical mutation. On the other hand, if mutations are too rapid, the pathogen evolution becomes volatile and, once again, it fails to spread. Between these two extremes, however, we identify a broad range of conditions in which an initially sub-pandemic pathogen can break through to gain widespread prevalence.
BibTeX
@misc{zhang2021epidemic,
  author = {Xiyun Zhang and Zhongyuan Ruan and Muhua Zheng and Jie Zhou and Stefano Boccaletti and Baruch Barzel},
  title = {{Epidemic spreading under mutually independent intra- and inter-host pathogen evolution}},
  year = {2021},
  eprint = {2102.11066},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2102.11066},
}
Figure