Early signs of infection help predict future spread of the disease

Most interspecies "spillovers" of viruses end in nothing: an individual animal (or several) gets infected, the chain breaks - and that's it. Only occasionally does the introduction lead to long-term circulation in a new population and large outbreaks. A team from Penn State demonstrated a simple but practical idea on an experimental model: early epidemiological signs immediately after a spillover can be used to estimate the chance that the virus will remain at the population level. In other words, not only the properties of the virus and the "donor" host are important - it is important how exactly the very first episode in the new host goes: how many individuals are infected, how often they shed the virus, and how vulnerable the host species is. These parameters, recorded "from the threshold", explain a significant share of the subsequent fate of the pathogen.

Background of the study

When a virus "jumps" to a new host species (spillover), its further fate is decided in a matter of "generations": the chain either dies out due to accidents and rare contacts, or it takes hold and becomes steadily transmitted. At this point, not only the biology of the virus works, but also the "small-scale epidemiology" of the start: how many individuals are infected at once, how often they actually shed the pathogen (shedding), how vulnerable the new species is. Classical stochastic epidemiology has long shown that random extinctions of foci are common in small numbers, and the success of the introduction is increased by the effects of "propagule pressure" - more sources at the start, a higher chance of not dying out.

The problem is that most real spillover events in wild animals are recorded late and irregularly: it is difficult to measure the earliest parameters. Therefore, laboratory systems are valuable, where interspecies "jumps" can be reproduced and early metrics can be measured in doses. Such a platform was the pair Orsay virus ↔ nematode Caenorhabditis: this is a natural RNA virus of the intestine of C. elegans, and related species differ in susceptibility and transmission - an ideal stand to separate "intra-host" barriers from "inter-host" ones. It was previously shown that the host spectrum of Orsay is wide, but heterogeneous - this is what empirical models of spillover and fixation are built on.

A new paper in PLOS Biology puts this idea into a rigorous experiment: the researchers induce the introduction of the virus into several “non-native” species, measure the prevalence of infection and the probability of shedding immediately after introduction, and then test whether the virus will persist in the population through a series of passages. It is these early epidemic signs – the breadth of coverage and the proportion of truly infectious individuals – that turn out to be the best predictors of subsequent success, while the “depth” of infection in individual carriers (viral load) predicts the outcome worse. This agrees well with mechanistic estimates of the probability of “not fading out” at each transplant and with the theory of stochastic burnout of outbreaks.

The practical implication for biosurveillance is simple: in addition to the characteristics of the pathogen itself and the reservoir species, early field investigations should assess two “fast” metrics in the recipient population as early as possible – how many are infected and who is actually infectious. These observables provide an informative “alarm signal” about the chances of establishment and help prioritize monitoring and containment resources before an outbreak develops.

How the hypothesis was tested: "nematode virus" and multiple passages

The authors used the well-studied Orsay virus ↔ Caenorhabditis nematode system: a naturally occurring RNA virus of the intestinal cells of C. elegans that is transmitted via the fecal-oral route and causes a mild, reversible infection - an ideal setup in which to repeatedly and reproducibly reproduce "jumps" between closely related species. The researchers induced spillover in eight strains belonging to seven "non-native" species for the virus, measured the prevalence of infection and the frequency of "shedding" of the virus (through co-culture with fluorescent "sentinels"), and then transferred small groups of adult worms to "clean" plates ten times in a row. If the virus continued to appear in PCR, it was "maintained" (held) in the new population; if the signal disappeared, it was lost. This protocol models the real spillover dilemma: can a pathogen overcome bottlenecks - from replication in new hosts to their infectivity - and avoid random extinction in the first generations?

What turned out to be the main "early clues"

In the "correlative" models, the number of passages before virus loss (simply: how long it persisted) was higher where immediately after introduction there was (1) a higher proportion of infected individuals (prevalence), (2) a higher probability that infected individuals actually shed virus (shedding), and (3) a higher relative susceptibility of the host species; however, the intensity of infection within an individual host (Ct in infected individuals) showed no significant relationship. When all indicators were included in one model, the first two - prevalence and shedding - were reliably "persistent", and together they explained more than half of the variation in outcome. This is an important practical conclusion: the breadth of coverage and infectivity at the start are more important than the "depth" of infection in each individual.

"Mechanistic" test: how many infectious people are needed for the transmission to take place

To go beyond correlations, the authors built a mechanistic model: using early measured metrics, they calculated the probability that at least one sufficiently infectious worm would end up on a new plate during the next transfer and “keep the fire” of transmission going. This mechanistic estimate alone explained ≈38% of the observed variation; adding prevalence, intensity, and random strain/experimental series effects increased the accuracy to ≈66%. That is, the basic epidemic “physics” of transmission already explains a lot, and early observed metrics add a significant amount of predictability.

Key figures of the experiment

In a series of four independent "blocks", the authors maintained 16 viral lines for each strain. In total, 15 lines in nematodes "non-native" to the virus survived all 10 passages with reliable detection of Orsay RNA by RT-qPCR, i.e. the virus gained a foothold; the rest dropped out earlier. Interestingly, of these "surviving" lines, 12 were in Caenorhabditis sulstoni SB454, two in C. latens JU724, and one in C. wallacei JU1873 - a clear example of how species susceptibility affects the chances of gaining a foothold even in very close hosts. "Biodosimetry" was used to calibrate susceptibility (TCID50/μl for each strain based on the highly sensitive control C. elegans JU1580).

Why this changes the focus of spillover monitoring

After high-profile zoonotic outbreaks (from Ebola to SARS-CoV-2), the response logic is often to ramp up surveillance where transmission is already visible. The new work adds a tool for very early triage of events: if we see a high proportion of infected people at the start, and infected people regularly “shine” as sources (shedding), this is a signal that the chance of the pathogen gaining a foothold is high, and such episodes require priority resources (from field trapping and sequencing to restrictive measures). But a high viral load in individuals without a wide prevalence is not a reliable predictor of population success.

How it was done technically (and why the result can be trusted)

The sentinel system helped to experimentally "sort out" the early signs: five transgenic reporter worms ( pals-5p::GFP ) were added to 15 "shedding candidates", and the glow for 3-5 days recorded the fact of transmission - a simple and sensitive benchmark of infectivity. Prevalence and intensity were calculated by RT-qPCR in small bullets (from a single worm to triplets), which works equally well at low and high proportions. Next, the "correlative" and "mechanistic" layers were combined in statistical models with random effects of strain, line, and passage number. Such "stitching" increases the transferability of results beyond a specific model and reduces the risk of "recalibrating" conclusions for a single system.

What this means for 'big' pathogens - cautious conclusions

Yes, the work was done on nematodes, not mammals. But the principles demonstrated are general: to gain a foothold after a spillover, a pathogen needs enough sources of infection and enough contacts already in the first steps; if these "units of infectivity" are few, stochastics quickly extinguish the outbreak (classic "Allais effects" and "propagule pressure"). Hence the practical heuristic: in early field investigations (be it bat viruses, bird flu or new host plants of phytopathogens), it is useful to prioritize rapid estimates of prevalence and shedding in the recipient population, and not rely only on the properties of the virus itself and its "donor" reservoir.

Where to Go Next: Three Directions for Research and Practice

• Field early metrics. Standardize “rapid” prevalence and shedding measurements (from traces, exometabolites, PCR/isotope traps) immediately after the first spillover signals - and test their predictive value in wild systems.
• Contact indicators. Integrate data on the frequency and structure of contacts in a new recipient population (density, mixing, migrations) into mechanistic assessments as a next step beyond "micro" metrics.
• Translation to zoonoses. Pilot protocols for trapping and screening for “early signs” in mammals/birds in known spillover hotspots, followed by post-hoc validation of whether the pathogen has become established or not.

Briefly - the main thing

• Early "broad" signs are more important than "deep" ones: high prevalence and virus shedding immediately after introduction are better predictors of population retention than the intensity of infection in individual carriers.
• The mechanistic model explains ≈38% of the variation in outcome using early data alone; with prevalence/intensity and random effects added, ≈66%.
• Monitoring practice: Record “who is infected” and “who is actually infecting” as early as possible - this helps to quickly understand where to direct resources so as not to miss the real risk.

Research source: Clara L. Shaw, David A. Kennedy. Early epidemiological characteristics explain the chance of population-level virus persistence following spillover events. PLOS Biology, August 21, 2025. https://doi.org/10.1371/journal.pbio.3003315