Spillover

by David Quammen

  An interesting thing about Hendra, Plowright told me, is that it’s one of four new viruses that emerged around the same time from this single group of bats, the pteropids. Soon after Hendra virus made its debut north of Brisbane, in 1994, there was Australian bat lyssavirus, appearing at two other sites along the Queensland coast, in 1996; then Menangle virus, emerging near Sydney, in 1997; and then Nipah virus, up in Malaysia, in September 1998. “For four viruses to emerge from one host genus within a short period of time is unprecedented,” she said. “So we feel there’s been some change in the ecology of Pteropus species that could precipitate disease emergence.” Hume Field had helped identify such contributing factors in the case of Nipah virus among the pig farms of Malaysia. Now, eight years later, with Field on her committee of dissertation advisers, Plowright was looking for similar factors in the matter of Hendra. Changes in habitat, she knew, had affected population size, distributional patterns, and migratory behavior of Hendra reservoir hosts—not just the little red flying fox but also its congenerics, the black flying fox, the grey-headed, and the spectacled. Her task was to investigate how those changes had affected in turn the distribution, prevalence, and spillover likelihood of the virus.

  Plowright’s project, like much work in ecology these days, entailed a combination of data-gathering from the field and mathematical modeling by computer. The basic conceptual framework, she explained, “was developed by two guys in the 1920s, Kermack and McKendrick.” She meant the SIR model (susceptible-infected-recovered), which I described earlier. Having alluded to the intellectual heritage, she began talking about susceptible individuals, infected individuals, and recovered individuals in a given bat population. If the population is isolated and insufficiently large, then the virus will move through it, infecting the susceptibles and leaving them recovered (and immune to reinfection), until there are virtually no susceptibles left. Then it will die out, just as measles dies out in an isolated human village. Eventually the virus will return, brought back to that population by a wayward, infected bat. This represents the same blinking-Christmas-light pattern that I invoked with regard to Marburg. The ecologists call it a metapopulation: a population of populations. The virus avoids extinction by infecting one relatively isolated population of bats after another. It dies out here, it arrives and infects there; it may not be permanently present in any one population but it’s always somewhere. The lights blink off/on in their turns, never all lit, never all dark. If the bat populations are separated by distances great enough that those distances are seldom crossed, then the rate of reinfection is slow. The lights blink off and on languidly.

  Now imagine one such bat population within the metapopulation. It has progressed through the SIR sequence, every individual infected, every one recovered, and the virus is gone. But not gone forever. As years pass, as the birth of new bats and the death of old ones raise back the proportion of susceptibles, the population regains its collective vulnerability to the virus. Greater isolation means greater elapsed time before the virus returns; greater elapsed time yields more newborn susceptibles; more susceptibles mean a richer potential for explosive infection. “So when you do introduce the virus again,” Plowright said, describing the godlike role of the modeler, “you get a much bigger outbreak.” This is where the Christmas-light metaphor fails to serve, because one light suddenly glows like a supernova among ordinary stars.

  Plowright of course was working with numbers, not analogies. But her numbers reflected roughly this scenario. The relevance of such modeling to the facts on the ground is that Australian populations of flying foxes have become more isolated in recent decades. “The east coast of Australia used to be one big contiguous forest,” she told me, “and so you had bat populations pretty evenly dispersed along the coastline.” Their roosting aggregations, in the old days, were relatively mobile. Their food resources—mainly nectar and fruit—were diverse, seasonally variable, and scattered patchily throughout the forest. Each group of bats, comprising maybe a few hundred or a few thousand individuals, would fly out to a feeding site at night, return at daylight, and also migrate seasonally to put themselves closer to concentrations of food. With all the coming and going, individual bats would sometimes transfer from one group to another, carrying Hendra virus with them if they happened to be infected. There was a continual mixing and a continual reinfection of the smallish groups. This seems to have been the situation—for the little red flying fox, for the other flying foxes, and for Hendra virus—from time immemorial. Then things changed.

  Habitat alteration was an ancient tradition in Australia, in the form of burning by Aboriginal people, but in recent decades land clearance has become a more drastic and mechanized trend, with less-reversible results, especially in Queensland. Vast areas of old forest have been cut, or chained down with bulldozers, to make way for cattle ranching and urban sprawl. People have planted orchards, established urban parks, landscaped their yards with blossoming trees, and created other unintended enticements amid the cities and suburbs. “So bats have decided that, as their native habitat is disappearing, as climate is becoming more variable, and their food source is becoming less diverse, it’s easier to live in an urban area.” They gather now in larger aggregations, traveling shorter distances to feed, living at closer proximity to humans (and to the horses that humans keep). Flying foxes in Sydney, flying foxes in Melbourne, flying foxes in Cairns. Flying foxes in the Moreton Bay fig trees shading a paddock on the north side of Brisbane.

  I saw where Plowright was going and tried to frame the last bit in my own mind. So those large aggregations—comprising bats that are more sedentary, more urban, less needful of flying long distances in search of wild food—tend to reinfect one another less frequently? And in the interim they accumulate more susceptible individuals? So when the virus does arrive, the spread of new infections is more sudden and intense? The virus is more prevalent and abundant?

  “Exactly. That’s it,” she said.

  “And then a great likelihood of spillover into another species?” I wanted to leap toward that easy epiphany but Plowright, with many bats yet to catch, many data to assemble, many model parameters to explore, held me back. Five years after our conversation, with the PhD finished, now a respected voice on Hendra herself, she would present her work and ideas in an august journal, the Proceedings of the Royal Society. But for the moment, amid the rains and high waters of the Northern Territory, she spoke provisionally.

  “This is a theory,” she said.

  82

  Theories require testing, as Raina Plowright well knew. Science proceeds by observation and supposition and testing. Another such supposition pertains to ebolaviruses. If you’ve been paying close attention, you’ll have noticed that just a few pages back I lumped Ebola virus, along with Hendra and Nipah and others, among viruses for which bats serve as reservoirs. So to clarify: That inclusion is tentative. It’s a hypothesis awaiting assessment against further evidence. No one, as of this writing, has isolated any live ebolavirus from a bat—and virus isolation is still the gold standard for identifying a reservoir. That may happen soon; people are trying. Meanwhile the Ebola-in-bats hypothesis seems stronger since Jonathan Towner’s team achieved their isolations of Marburg virus, so closely related, also from bats. And it has been strengthened further, at least a little, by another bit of data added to the ebolavirus dossier about the same time. This bit came in the form of a story about a little girl.

  Eric Leroy, the Paris-trained virologist based in Franceville, Gabon, who had been chasing Ebola for more than a decade, led the team that reconstructed the girl’s story. Their new evidence derived not from molecular virology but from old-fashioned epidemiological detective work—interviewing survivors, tracing contacts, discerning patterns. The context was an outbreak of Ebola virus that occurred in and around a village called Luebo, along the Lulua River, in a southern province of the Democratic Republic of the Congo. Between late May and November 2007, more than 260 people sickened with wh
at seemed to be or (in some confirmed cases) definitely was Ebola virus. Most of them died. The lethality was 70 percent. Leroy and his colleagues arrived in October, as part of an international WHO response team in cooperation with the DRC’s Ministry of Health. Leroy’s study focused on the network of transmissions, which all seemed traceable to a certain fifty-five-year-old woman. She became known, in their report, as patient A. She wasn’t necessarily the first human to get infected; she was merely the first identified. This woman, elderly by Congo village standards, died after suffering high fever, vomiting, diarrhea, and hemorrhages. Eleven of her close contacts, mainly family, who helped care for her, sickened and died too. The outbreak spread onward from there.

  Leroy and his group wondered how the woman herself had gotten infected. No one in her village showed symptoms before she did. So the investigators broadened their search to surrounding villages, of which there were quite a few, both along the river and in the forest nearby. From their interviews and their legwork, they learned that the villages were interconnected by footpaths, and that on Mondays the heavy traffic led to one particular village, Mombo Mounene 2, the site of a big weekly market. They also learned about an annual aggregation of migrating bats.

  The bats generally arrived in April and May, stopping over amid a longer journey, finding roost sites and wild fruit trees on two islands in the river. In an average year, there might be thousands or tens of thousands of animals, according to what Leroy’s group heard. In 2007, the migration was especially large. From their island roosts, the bats ranged the area. Sometimes they fed at a palm oil plantation along the river’s north bank; the plantation was a leftover from colonial times, now abandoned and gone derelict, but still offering palm fruits in April on its remaining trees. Many or most of the animals were hammer-headed fruit bats (Hypsignathus monstrosus) and Franquet’s epauletted fruit bats (Epomops franqueti), two of the three in which Leroy had earlier found Ebola antibodies. While roosting, the bats dangled thickly on tree branches. Local people, hungry for protein or a little extra cash, hunted them with guns. Hammer-headed bats, big and meaty, were especially prized. A single shotgun blast could bring down several dozen bats. Many of those animals ended up, freshly killed, raw and bloody, in the weekly market at Mombo Mounene 2, from which buyers carried them home for dinner.

  One man who regularly walked from his own village to the market, and often bought bats, seems to have suffered a mild case of Ebola. The investigators eventually labeled him patient C. He wasn’t a bat hunter himself; he was a retail consumer. During late May or early June, according to patient C’s own recollection, he weathered some minor symptoms, mainly fever and headache. He recovered, but that wasn’t the end of it. “Patient C was the father of a 4-year-old girl (patient B),” Leroy and his team later reported, “who suddenly fell ill on 12 June and died on 16 June 2007, having had vomiting, diarrhoea, and high fever.” The little girl didn’t hemorrhage, and she was never tested for Ebola, but it’s the most plausible diagnosis.

  How had she contracted it? Possibly she had shared in eating a fruit bat that carried the virus. What are the odds faced by bat-eaters? Hard to say; hard even to guess. If the hammer-headed bat is an Ebola reservoir, what’s the prevalence of the virus within a given population? That’s another unknown. Towner found 5 percent prevalence of Marburg in Egyptian fruit bats, meaning that one animal in twenty could be infected. Assuming a roughly similar prevalence in the hammer-headed bat, the little girl’s family had been unlucky as well as hungry. They might have eaten nineteen other bats and gotten no exposure. Then again, if a bat meal was shared, why didn’t the girl’s mother and other family members get sick? Possibly her father, infected or besmeared after purchasing bats in the market, had carried the girl (common practice with small children thereabouts) along the footpath back to their village. The father, patient C, seems to have passed the virus to nobody else.

  But his little daughter did pass it along. Her dead body was washed for burial, in accord with local traditions, by a close friend of the family. That friend was the fifty-five-year-old woman who became patient A.

  “Thus, virus transmission may have occurred when patient A prepared the corpse for burial ceremony,” Leroy’s group wrote. “When interviewed, the two other preparers, the girl’s mother and grandmother, reported they did not have direct contact with the corpse and they did not develop any clinical sign of infection in the four following weeks.” Their role in the funerary washing was apparently observational. They didn’t touch the dead body of their daughter and granddaughter. But patient A did, performing faithfully the service of a close family friend, after which she went back to her life—what was left of it. She resumed her social interactions, and 183 other people caught Ebola and died.

  Leroy’s team reconstructed this story and then, keen to extract meaning, asked themselves several questions. Why had the father infected his daughter but no one else? Maybe because he had a mild case, with a low level of virus in his body and not much leaking out. But if his case was mild, why was his daughter’s so severe, killing her within four days? Maybe because, as a small child racked with vomiting and diarrhea, she had died of untreated dehydration. Why was there only one bat-to-human spillover event? Why was patient C unique, as the sole case linked directly to the reservoir? Well, maybe he wasn’t. He was just the only one that came to notice. “In fact, it is highly likely that several other persons were infected by bats,” Leroy’s group wrote, “but the circumstances required for subsequent human-to-human transmission were not present.” They were alluding to dead-end infections. A person sickens, suffers solitarily or with carefully distanced succor from wary family or friends (food and water left at the door of a hut), and dies. Is buried unceremoniously. Eric Leroy didn’t know how many unfortunate people in the Luebo area may have eaten a bat, touched a bat, become infected with Ebola, succumbed to it, and been dropped into a hole, having infected no one else. Amid the horrific confusion of the outbreak, in those remote villages, the number of such dead-end cases might have been sizable.

  This brought Leroy’s team to the pivotal question. If the circumstances required for human-to-human transmission hadn’t been met, what were those circumstances? Why hadn’t the Luebo outbreak gone really big? Why hadn’t the tinder ignited the logs? It had started in May, after all, and WHO didn’t get there until October.

  83

  Human-to-human transmission is the crux. That capacity is what separates a bizarre, awful, localized, intermittent, and mysterious disease (such as Ebola) from a global pandemic. Remember the simple equation offered by Roy Anderson and Robert May for the dynamics of an unfolding epidemic?

  R0 = βN/(α + b + v)

  In that formulation, β represents the transmission rate. β is the letter beta, in case you’re not a mathematician or a Greek. Here it’s a multiplier in the single expression that stands as numerator of the fraction, a strong position. What that means is, when β changes muchly, R0 changes muchly. And R0, your good memory tells you, is the measure of whether an outbreak will take off.

  In some zoonotic pathogens, efficient transmissibility among humans seems to be inherent from the start, a sort of accidental preadaptedness for spreading through the human population, despite a long history of residence within some other host. SARS-CoV had it, from the earliest days of its 2002–2003 emergence in Guangdong and Hong Kong. SARS-CoV has it, no matter where or why SARS-CoV may be hiding since then. Hendra virus does not have it. Hendra achieves fluent transmission among horses but not among humans. Of course, a pathogen may also acquire that capacity by mutation and adaptation within human hosts. Have you noticed the persistent, low-level buzz about avian influenza, the strain known as H5N1, among disease experts over the past fifteen years? That’s because avian flu worries them deeply, though it hasn’t caused many human fatalities. Swine flu comes and goes periodically in the human population (as it came and went during 2009), sometimes causing a bad pandemic and sometimes (as in 2009) not so bad as ex
pected; but avian flu resides in a different category of menacing possibility. It worries the flu scientists because they know that H5N1 influenza is (1) extremely virulent in people, with a high lethality though a relatively low number of cases, and yet (2) poorly transmissible, so far, from human to human. It’ll kill you if you catch it, very likely, but you’re unlikely to catch it except by butchering an infected chicken. Most of us don’t butcher our own chickens, and health officials all over the world have been working hard to assure that the chickens we handle—dead, disarticulated, wrapped in plastic or otherwise—have not been infected. But if H5N1 mutates or reassembles itself in just the right way, if it adapts for human-to-human transmission, then H5N1 could become the biggest and fastest killer disease since 1918.

  How does a pathogen acquire such an adaptation? The process of genetic variation (by mutation or other means) is random. A game of craps. But an abundance of opportunity helps to increase a virus’s likelihood of rolling its point—that is, chancing into a highly adaptive change. The more rolls before sevening out, the more opportunities to win. And there’s Jon Epstein’s word again: opportunity.