杏吧原创

鈥淲HEN acorns fill ye olde oak trees, t鈥檞ill be a bad year for Lyme disease.鈥�
Folklore and disease control make strange bedfellows but this could be the
future of epidemiology. Along with doom-filled doggerel you may also find,
posted at the entrance to your local wood, the words: 鈥淗igh risk of Lyme
disease, enter at your peril.鈥� Perhaps your new home will come with a public
health warning: 鈥淟yme disease has been identified as a hazard in this area. All
gardens must be sprayed annually with tick-killing insecticides.鈥�

This all seems rather far-fetched in countries like Britain where Lyme
disease is still uncommon, but data from the US show that the situation can
change in little more than a decade. American figures record 500 cases in 1982.
Last year there were more than 16 000. Lyme disease is now the most common
vector-borne disease in the US. It is part of a wider problem for which we
ourselves are to blame. Throughout the world, humans have altered the landscape
and moved into new environments, crossing paths with all sorts of animals and
their exotic diseases鈥攏ot just Lyme disease, but Ebola virus and
hantavirus infections and old enemies such as plague and cholera
(see 鈥淭esting the waters鈥�).
Our changing lifestyles have put us in contact with these
pathogens; now epidemiologists want to change the way we deal with them, and
Lyme disease is in the vanguard.

鈥淢any of these diseases are so complex that it takes a different approach to
try to understand them,鈥� says Uriel Kitron at the University of Illinois at
Urbana-Champaign. He is a proponent of landscape epidemiology, a discipline that
aims to identify and predict disease hot spots by combining traditional
ecological fieldwork and new methods such as remote sensing, molecular detection
of pathogens and computer modelling. The new approach has already acquired an
impressive following. Researchers at the US Centers for Disease Control and
Prevention (CDC) are using it to draw up a national map of Lyme disease risk and
to investigate interactions between the plague bacterium and the environment.
NASA scientists too are putting landscape epidemiology into practice as they use
satellite data to study Lyme disease, along with cholera, leishmaniasis,
Venezuelan equine encephalitis and malaria at the Center for Health Applications
of Aerospace Related Technologies (CHAART) in Moffett Field, California.

Epidemiologists have traditionally mapped the prevalence of diseases around
the globe using information about the number of sick people resident at each
location. This has its drawbacks. Such a map, for example, would show 529 cases
of Lyme disease in Ohio since 1989 even though fewer than a dozen specimens of
deer ticks have ever been found in that state. The alternative, tracking disease
organisms down to their source, is essential if the information is to be used in
disease prevention. But pinpointing pathogens is no simple task. It involves
identifying the habitats preferred by the animals that serve as disease
reservoirs and by the vectors, and then working out what triggers boom or bust
in these populations. The landscape epidemiologists also need accurate maps that
distinguish between different habitats, plus up-to-date information about
precipitation, temperature and other variables that might affect disease
risk.

Lyme disease, which is currently the primary focus for the landscape
epidemiology approach, is a good example. White-footed mice and other small
rodents are the most common disease reservoirs for the spirochete bacterium
Borrelia burgdorferi that causes the disease, but it can also live in
lizards and birds. But these animals cannot pass the bacterium to humans
directly. Instead it must pass through an intermediary, ticks of the genus
Ixodes. There are two species of tick that transmit the spirochete to
humans in Europe and Asia, and two in North America. For part of the year, the
ticks require large animals to survive. In the US, and possibly Europe too, deer
are a preferred host, and as they roam the countryside, they take Lyme disease
with them. In the US the worst affected areas are the Northeast, the Upper
Midwest and the West Coast. Where exactly the disease is lurking, though,
depends on environmental variables such as vegetation and topography, which
affect the distribution of all the creatures in the disease cycle. On a local
scale these factors come together only in certain patches. These are the
high-risk areas scientists hope to home in on.

But where to begin the search? History can teach researchers a lesson or two.
鈥淭hese ticks were rare species 20 years ago,鈥� says Durland Fish, a medical
entomologist at Yale University. Since then there has been a population boom in
white-tailed deer, as predators such as wolves and mountain lions have declined
and forest cover has increased. White-footed mice are also thriving, especially
in disturbed habitats and oak forests in the eastern US that were planted this
century to replace native chestnuts wiped out by a fungal disease. 鈥淣ow ticks
have expanded their range and increased their density in the Northeast and
Midwest to perhaps the highest ever in history,鈥� says Fish.

鈥淲e鈥檝e done a lot to increase the risk of Lyme disease,鈥� says Richard
Ostfeld, an ecologist at the Institute of Ecosystem Studies in Millbrook, New
York, who believes that the degradation of forest habitats is a major factor.
Working with Josh Van Buskirk, now at the University of Zurich, Ostfeld has
produced a computer model which shows that the more diverse an ecosystem is, the
lower the risk of Lyme disease. In a rich ecosystem, the researchers say, ticks
feed on a wider variety of hosts鈥攕uch as birds and lizards鈥攚hich
don鈥檛 sustain the Lyme spirochete as well or for as long as white-footed
mice.

Evidence from the field supports this claim. Several groups of researchers
have been using genetic tests to estimate the proportion of ticks that carry the
Lyme disease spirochete in various parts of the country. They have discovered
that in the simpler ecosystems of the Northeast and Upper Midwest, between 15
and 25 per cent of nymphal ticks are infected, whereas in the more complex
ecosystems of southern and western regions of the US only 1 to 5 per cent carry
the spirochete.

Another clue to disease hot spots comes from the incidence of Lyme disease
itself, which varies greatly from year to year at various locations. Last year,
for instance, there were 16 461 cases in the US as a whole鈥攗p 40 per cent
on 1995. Northeastern states were particularly hard hit. And Ostfeld believes he
knows why.

His team at the Institute of Ecosystem Studies has identified acorns as a
driving factor in disease prevalence. For five summers, starting in 1991, they
counted ticks in the various habitat types of semirural New York. Oak and maple
forests contained most ticks, with highs alternating between the two habitats.
Tick numbers peaked in the oak forests the summer after good acorn years and in
maple forests the summer following bad acorn years. Ostfeld and his team
believed that the acorns were attracting animals involved in the Lyme disease
cycle. To test their hypothesis they added a million or so acorns to several
forest plots. Sure enough, the following summer the populations of the tick
larvae were around 10 times as high in the test sites as in control areas.

The team also documented the two-year cycle of events following bumper acorn
years such as 1991 and 1994. First, deer were attracted into the oak forests,
where the adult ticks riding on them would drop off, mate and lay eggs. More
importantly, says Ostfeld, hordes of mice also moved in, and the abundance of
acorns allowed large numbers of them to survive the winter. So the next summer
there were peak populations both of larval ticks and of mice鈥攎any infected
with the Lyme spirochete鈥攆or the ticks to feed on. During the following
spring and summer the oak forest erupted with a bounty of infected nymphal
ticks, the main transmitters of Lyme disease to humans. The nymphs, unnoticeable
because of their pinpoint size, latched on to hapless hikers for a meal of blood
and injected them with the Lyme disease spirochete.

Forecasting hot spots

A large acorn crop in 1994 led Ostfeld鈥檚 team to predict that 1996 would be a
big year for Lyme disease. 鈥淎nd in our county, Duchess County, New York, it
was,鈥� he says. The researchers have had less success forecasting disease hot
spots in maple forests, but they believe that the acorn connection has
far-reaching implications. It could be used to predict high-risk years for Lyme
disease wherever there are large areas of oak-dominated forest, including the US
West Coast, Europe and western Asia. Acorn production is affected by weather and
synchronised in trees across hundreds of kilometres. By monitoring natural
fluctuations in the crop size, researchers should be able to give two years鈥�
warning of a disease outbreak. Forest rangers could then steer hikers away.

But this idea is not universally accepted. In Rhode Island, tick populations
fluctuate as much as two and three-fold from year to year. There, Thomas Mather,
an entomologist at the University of Rhode Island in Kingston, and his team have
spent the past five years counting ticks at 80 forested sites. They can find
only one explanation for the variability鈥攑recipitation. 鈥淣ymphal tick
populations have swelled when both winter snow and spring precipitation have
been high, and so have the number of cases of Lyme disease,鈥� says Mather.
Moisture is needed to prevent the larval ticks from drying out as they moult
into nymphs and overwinter in the forest litter, he explains. So climatic
conditions greatly affect their survival.

鈥淭his past year we had no snow cover and little rain during April, May and
early June,鈥� says Mather. 鈥淚 went out on a limb and told the local press to
expect a low tick year in 1997 and now that the season has ended, it seems I was
right. Populations were down about 50 per cent from last year.鈥� Now he is
hopeful that some signals in weather patterns could be monitored on a large
scale using remote sensing data collected by satellites so that Lyme disease
outbreaks could be forecast.

Satellite data have already been used to help map the high-risk areas for
Lyme disease in Westchester County, New York鈥攁 suburban area hit hard by
the disease. Fish collaborated with researchers at CHAART to pinpoint the
neighbourhoods where ticks cluster. First his team enlisted local veterinarians
to get blood samples from 2000 pet dogs which were then tested for the presence
of antibodies indicating exposure to the Lyme disease spirochete. Then the group
mapped where the dogs lived and overlaid it with a map of local vegetation
generated from LANDSAT satellite data. They found that many of the dogs exposed
to the spirochete lived on properties with dense vegetation, close to forest.
The results were not unexpected, but still exciting. 鈥淭his was the first time
anyone had used satellite imagery to predict the activity of a pathogen,鈥� says
Fish. It means that in future, researchers may be able to map Lyme disease risk
across the nation without relying on laborious tick-counting surveys or victims鈥�
guesses about where they were bitten.

Taking their investigation a step further, the researchers have shown that
LANDSAT measurements of greenness and wetness鈥攚hich vary by vegetation
type鈥攃an also predict where high-risk properties are located. 鈥淲e can look
up the address of each of 800 000 residents [in Westchester County], overlay a
satellite image, and that will tell us if it鈥檚 a high-risk property or not,鈥�
Fish says. To test their model they asked 1000 residents of a nearby county if
they had been bitten by a tick near their home in the past year. Fish and his
team had correctly identified the high-risk properties in 77 per cent of cases.
The study is due to be published in the December issue of The American
Journal of Tropical Medicine and Hygiene.

But this approach has severe limitations. Mather points out that satellite
measurements of greenness would label parts of northern Rhode Island as
high-risk, even though the area has no deer, no ticks and no Lyme disease. 鈥淚t鈥檚
still too early to know if we鈥檙e finding universal correlates,鈥� says Kitron, who
studies Lyme disease in Wisconsin. He says the disease ecology in the Midwest
appears to be a bit more complex, with chipmunks playing an important role as
hosts and with proximity to rivers and soil type affecting tick distribution. On
the West Coast, the other Lyme disease hot spot in the US, the disease ecology
is even more complex. The ticks have several hosts, not all of which can sustain
the spirochete, and there are several different tick species, of which only one
bites humans. Furthermore, fieldwork suggests that the high-risk areas are
hardwood forests with plenty of leaf litter, not the suburban residential areas
where Lyme disease often lurks in the Northeast.

On the map

So different kinds of data may be needed to estimate risk in each of the
Northeast, the Upper Midwest and the West, and the CDC鈥檚 goal of creating a
national Lyme disease risk map is quite a challenge. The CDC has just completed
a preliminary risk map, based on county-by-county reports of the presence of
ticks and disease cases. But the quality of the data is variable, says David T.
Dennis, who coordinates the CDC鈥檚 Lyme disease programme in Fort Collins,
Colorado. 鈥淚t鈥檚 a hodgepodge; no one has done this systematically,鈥� he says.

The agency hopes to improve the accuracy of the map鈥攁nd get around
doing laborious tick surveys鈥攂y identifying landscape features that
predict high tick numbers in various regions of the country. To start with, the
CDC is working with Mather and Fish to collect field data and search for
environmental correlates in the Northeast. 鈥淭his is going to be a long-term
project,鈥� says Dennis.

The complexity of all this has left some scientists wondering whether they
will succeed. 鈥淚 don鈥檛 know if we鈥檒l be any better than the weather predictors
in the 1960s or the earthquake predictors,鈥� says Gregory Gurri Glass, a disease
ecologist at Johns Hopkins University in Baltimore. But at least weather
forecasting has improved in the past three decades. And recent successes in
pinpointing Lyme disease have left many researchers optimistic about the
potential of landscape epidemiology too. 鈥淭o me, this is the future of
epidemiology,鈥� says Fish.

* * *

Testing the waters

WHICH satellite measurements should the new breed of epidemiologists be
watching? Those studying cholera may have to look no further than the sea
surface. During a recent examination of satellite data, NASA scientists
discovered an association between the height and temperature of the sea surface
and outbreaks of cholera in Bangladesh in 1992 and 1995.

The research is in collaboration with Rita Colwell and her colleagues at the
University of Maryland, who recognised that cholera cases in Bangladesh surge in
spring and autumn when coastal waters are at their warmest. Work in her
laboratory showed that cholera-carrying plankton thrive in these conditions.

Sea levels rise when waters warm up, so researchers at the University of
Maryland and the Center for Health Applications of Aerospace Related
Technologies (CHAART) in Moffett Field, California, examined satellite data on
surface water temperature and height. Sure enough, peaks in these measurements
corresponded to the major cholera outbreaks. If satellite monitoring of the sea
surface really can predict cholera, the method would give several weeks鈥� warning
of an outbreak. 鈥淭hen we could manage the outbreak better with more resources
and minimise the loss of human life,鈥� says Brad Lobitz, a research scientist at
CHAART.

Colwell and her team also believe that Peru鈥檚 devastating cholera outbreak of
1991 may have been linked to El Ni帽o, which brings warm waters from the
western Pacific to central and eastern areas of the ocean and affects the global
climate. A large El Ni帽o is due to peak in the next few months and an
international research programme is already assessing its impact on disease.
Epidemiologists are working with climatologists to assess how climatic
conditions affect the incidence of many vector-borne diseases, including
malaria, plague and hantavirus.