An outbreak of coronavirus infections in Washington State that began at the end of February was likely sparked by a traveler who entered the state in mid-February, rather than by an infected person who flew from Wuhan, China, to Seattle in mid-January, as was previously thought. The data point to Germany and Italy outbreaks starting independently by travelers coming from China and then transmission of the virus from Italy to New York in late February. That’s the conclusion of a preprint posted on bioRxiv May 23 that modeled various transmission scenarios based on the genomes of viral samples collected from patients in Washington State and elsewhere.

Initial studies by Trevor Bedford, who studies pathogen evolution at the Fred Hutchinson Cancer Research Center and the University of Washington, indicated that the viral genomes from the first case identified in Washington State in January, known as WA1, and a patient diagnosed in late February in the same state were similar enough to suggest there had been “cryptic transmission” for weeks before the outbreak was recognized. 

Armed with much more genomic information amassed in the ensuing months, Michael Worobey of the University of Arizona and his colleagues simulated 1,000 different viral evolution scenarios to see if any would connect WA1 to the Washington infections six weeks later. Yet the models didn’t spit out the later infections’ viral sequences. “[W]e failed to observe a single simulated epidemic that has the characteristics of the real phylogeny,” the authors write in their report. 

“In all likelihood this didn’t start with WA1,” Worobey tells STAT. “It started with some unidentified person who arrived in Washington state at some later point. And we don’t know from where.” 

The new study also examined the likely origins of Europe’s first outbreaks, concluding that a cluster of cases in Germany did not seed the pandemic in Italy. Instead, the data point to both countries’ outbreaks starting independently by travelers coming from China, and then transmission of the virus from Italy to New York in late February.

An alternative scenario in which both the Germany and Italy outbreaks were independently introduced from China is further supported when missing sequences from undersampled locations are explicitly accommodated in the phylogeny. Using the approach described above, the evolutionary relationships of BavPat1 and viruses from the Italian outbreak were reconstructed while accommodating unsampled viruses assigned to Hubei and other locations (including Italy) determined to be undersampled, based on incidence data.

The resulting phylogeny strongly supports independent viral introductions from Hubei into Germany and Italy, with the ancestral location states in Hubei supported by a posterior probability of 0.79.

Worobey and his authors note that because undetected cases in Washington were unlikely between mid-January and mid-February, the US missed opportunities for testing and contact tracing to prevent an outbreak.

“Our finding that the virus associated with the first known transmission network in the US did not enter the country until mid-February is sobering,” the authors write, “since it demonstrates that the window of opportunity to block sustained transmission of the virus stretched all the way until that point.”