University of Arizona
A new study combines evolutionary genomics
from coronavirus samples with computer-simulated epidemics and detailed travel
records to reconstruct the spread of coronavirus across the world in
unprecedented detail.
Published in the journal Science, the results suggest an extended
period of missed opportunity when intensive testing and contact tracing might
have prevented SARS-CoV-2 from becoming established in North America and
Europe.
The paper also challenges suggestions that linked the
earliest known cases of COVID-19 on each continent in January to outbreaks
detected weeks later, and provides valuable insights that could inform public
health response and help with anticipating and preventing future outbreaks of
COVID-19 and other zoonotic diseases.
"Our aspiration was to develop and apply powerful
new technology to conduct a definitive analysis of how the pandemic unfolded in
space and time, across the globe," said University of Arizona researcher
Michael Worobey, who led an interdisciplinary team of scientists from 13
research institutions in the U.S., Belgium, Canada and the U.K.
"Before,
there were lots of possibilities floating around in a mish-mash of science,
social media and an unprecedented number of preprint publications still
awaiting peer review."
The team based their analysis on results from viral
genome sequencing efforts, which began immediately after the virus was
identified. These efforts quickly grew into a worldwide effort unprecedented in
scale and pace and have yielded tens of thousands of genome sequences, publicly
available in databases.
Contrary to widespread narratives, the first documented
arrivals of infected individuals traveling from China to the U.S. and Europe
did not snowball into continental outbreaks, the researchers found.
Instead, swift and decisive measures aimed at tracing and
containing those initial incursions of the virus were successful and should
serve as model responses directing future actions and policies by governments
and public health agencies, the study's authors conclude.
A Chinese national flying into Seattle from Wuhan, China
on Jan. 15 became the first patient in the U.S. shown to be infected with the
novel coronavirus and the first to have a SARS-CoV-2 genome sequenced. This
patient was designated 'WA1.' It was not until six weeks later that several
additional cases were detected in Washington state.
"And while all that time goes past, everyone is in
the dark and wondering, 'What's happening?'" Worobey said. "We hope
we're OK, we hope there are no other cases, and then it becomes clear, from a
remarkable community viral sampling program in Seattle, that there are more
cases in Washington and they are genetically very similar to WA1's virus."
Worobey and his collaborators tested the prevailing
hypothesis suggesting that patient WA1 had established a transmission cluster
that went undetected for six weeks.
Although the genomes sampled in February and
March share similarities with WA1, they are different enough that the idea of
WA1 establishing the ensuing outbreak is very unlikely, they determined. The
researchers' findings indicate that the jump from China to the U.S. likely
occurred on or around Feb. 1 instead.
The results also puts to rest speculation that this
outbreak, the earliest substantial transmission cluster in the U.S., may have
been initiated indirectly by dispersal of the virus from China to British
Columbia, Canada, just north of Washington State, and then spread from Canada
to the U.S.
Multiple SARS-CoV-2 genomes published by the British Columbia
Center for Disease Control appeared to be ancestral to the viral variants
sampled in Washington State, strongly suggesting a Canadian origin of the U.S.
epidemic. However, the present study revealed sequencing errors in those
genomes, thus ruling out this scenario.
Instead, the new study implicates a direct-from-China
source of the U.S. outbreak, right around the time the U.S. administration implemented
a travel ban for travelers from China in early February. The nationality of the
"index case" of the U.S. outbreak cannot be known for certain because
tens of thousands of U.S. citizens and visa holders traveled from China to the
U.S. even after the ban took effect.
A similar scenario marks the first known introduction of
coronavirus into Europe.
On Jan. 20, an employee of an automotive supply
company in Bavaria, Germany, flew in for a business meeting from Shanghai,
China, unknowingly carrying the virus, ultimately leading to infection of 16
co-workers.
In that case, too, an impressive response of rapid testing and
isolation prevented the outbreak from spreading any further, the study
concludes. Contrary to speculation, this German outbreak was not the source of
the outbreak in Northern Italy that eventually spread widely across Europe and
eventually to New York City and the rest of the U.S.
The authors also show that this China-to-Italy-US
dispersal route ignited transmission clusters on the East Coast slightly later
in February than the China-to-US movement of the virus that established the
Washington State outbreak. The Washington transmission cluster also predated
small clusters of community transmission in February in California, making it
the earliest anywhere in North America.
Early Containment
Works
The authors say intensive interventions, involving
testing, contact tracing, isolation measures and a high degree of compliance of
infected individuals, who reported their symptoms to health authorities and
self-isolated in a timely manner, helped Germany and the Seattle area contain
those outbreaks in January.
"We believe that those measures resulted in a
situation where the first sparks could successfully be stamped out, preventing
further spread into the community," Worobey said.
"What this tells us
is that the measures taken in those cases are highly effective and should serve
as a blueprint for future responses to emerging diseases that have the
potential to escalate into worldwide pandemics."
To reconstruct the pandemic's unfolding, the scientists
ran computer programs that carefully simulated the epidemiology and evolution
of the virus, in other words, how SARS-CoV-2 spread and mutated over time.
"This allowed us to re-run the tape of how the epidemic
unfolded, over and over again, and then check the scenarios that emerge in the
simulations against the patterns we see in reality," Worobey said.
"In the Washington case, we can ask, 'What if that
patient WA1 who arrived in the U.S. on Jan. 15 really did start that outbreak?'
Well, if he did, and you re-run that epidemic over and over and over, and then
sample infected patients from that epidemic and evolve the virus in that way,
do you get a pattern that looks like what we see in reality? And the answer was
no," he said.
"If you seed that early Italian outbreak with the
one in Germany, do you see the pattern that you get in the evolutionary data?
And the answer, again, is no," he said.
"By re-running the introduction of SARS-CoV-2 into
the U.S. and Europe through simulations, we showed that it was very unlikely
that the first documented viral introductions into these locales led to
productive transmission clusters," said co-author Joel Wertheim of the
University of California, San Diego.
"Molecular epidemiological analyses
are incredibly powerful for revealing transmissions patterns of
SARS-CoV-2."
Other methods were then combined with the data from the
virtual epidemics, yielding exceptionally detailed and quantitative results.
"Fundamental to this work stands our new tool
combining detailed travel history information and phylogenetics, which produces
a sort of 'family tree' of how the different genomes of virus sampled from
infected individuals are related to each other," said co-author Marc Suchard
of the University of California, Los Angeles.
"The more accurate
evolutionary reconstructions from these tools provide a critical step to
understand how SARS-CoV-2 spread globally in such a short time."
"We have to keep in mind that we have studied only short-term
evolution of this virus, so it hasn't had much time to accumulate many
mutations," said co-author Philippe Lemey of the University of Leuven,
Belgium.
"Add to that the uneven sampling of genomes from different parts
of the world, and it becomes clear that there are huge benefits to be gained
from integrating various sources of information, combining genomic
reconstructions with complementary approaches like flight records and the total
number of COVID-19 cases in various global regions in January and
February."
"Our research shows that when you do early
intervention and detection well, it can have a massive impact, both on
preventing pandemics and controlling them once they progress," Worobey
said.
"While the epidemic eventually slipped through, there were early
victories that show us the way forward: Comprehensive testing and case
identification are powerful weapons."
Funding sources for this study include the David and
Lucile Packard Foundation, the National Institutes of Health, the European Research
Council, the Wellcome Trust and the Canadian Institutes of Health Research
Coronavirus Rapid Response Programme.
