Abstract
In June 2020, the Xinfadi (XFD) Wholesale Market in Beijing became the epicenter of a sudden COVID-19 outbreak, totaling 368 cases after nearly two months of zero local transmission. While genomic sequencing identified the virus as lineage B1.1.29—common in Europe at the time—the specific transmission route remained debated. This study applies "leaf-dating" molecular clock analysis to SARS-CoV-2 genomes from the outbreak. Our analysis reveals that the XFD genomes exhibited a significant lack of genetic divergence, with expected collection dates months earlier than their actual recorded dates. This evolutionary stasis strongly suggests the virus was preserved in a frozen state before entering the market. These findings provide molecular evidence that contaminated cold-chain imports triggered the re-emergence, highlighting the need for expanded viral surveillance in international food trade.
Introduction
After 55 days without a single local case, Beijing faced a resurgence of COVID-19 in June 2020. The outbreak was centered at the Xinfadi (XFD) Wholesale Market, a critical hub for the city's food supply. Genetic tracing quickly categorized the virus as part of the B1.1.29 lineage, but the source of the introduction was unclear. Two primary hypotheses emerged: either an undetected human carrier had introduced the virus to the city, or the virus had arrived via contaminated frozen goods. Distinguishing between these possibilities is vital for public health strategy, as the latter implies a "cold-chain" transmission route that requires fundamentally different monitoring and quarantine protocols than human-to-human transmission.
Methodology: The Molecular Clock and Leaf-Dating
To investigate the origin of the XFD outbreak, researchers utilized a technique known as leaf-dating. This Bayesian phylogenetic approach relies on the concept of the molecular clock—the observation that viruses accumulate mutations at a relatively constant rate over time. By measuring the genetic divergence of a sample, scientists can estimate its "biological age" or expected collection date.
If a virus is circulating naturally in a human population, it will continue to mutate. However, if a virus is frozen, its evolutionary process is effectively paused. The researchers used BEAST2 software, employing a HKY85 nucleotide substitution model, to compare the recorded collection dates of the XFD samples against their expected dates based on their genetic profiles. They quantified the results using Bayes factors (BF) to test the hypothesis that the expected collection date differed significantly from the recorded date (H2) against the null hypothesis that they matched (H1).
Datasets and Controls
The study analyzed several genomic datasets sourced from GISAID to ensure a robust comparison:
- XFD Genomes: Three human-source SARS-CoV-2 genomes collected at the start of the Beijing outbreak (June 2020).
- Domestic Dataset: 241 genomes collected across mainland China (Dec 2019 – July 2020) to establish a baseline mutation rate.
- Lineage-Specified Dataset: 215 genomes from the B1.1.29 lineage to provide specific evolutionary context.
- Control Genomes: Five B1.1.29 genomes from other global regions, also collected on June 11, 2020, used to validate that the leaf-dating method would correctly identify "current" (non-frozen) viruses.
Key Results
Evolutionary Stasis
The analysis revealed a striking discrepancy in the XFD samples. The genomes were "too young" for their recorded age, lacking the mutations that should have accumulated between March and June 2020.
Specifically, using the domestic dataset as a baseline, the XFD genome IVDC-02-06 was estimated to have an expected collection date of March 11, 2020—92 days earlier than its actual collection on June 11. The Bayes factor for this discrepancy was 11.27, indicating "strong evidence" that the virus had not been evolving normally. The other two XFD genomes showed similar gaps of 68 and 49 days.
The Singapore Connection
Further supporting the "frozen" hypothesis, the researchers identified a real-world genetic match. One XFD genome was found to be genetically identical to a sample collected from a traveler in Singapore on March 26, 2020. This suggests that the virus likely entered a state of preservation (storage) around that time before reappearing in Beijing months later.
Validation via Controls
To ensure the model wasn't simply biased toward earlier dates, the researchers ran the same analysis on the five control genomes. In all control cases, the Bayes factors favored the recorded dates (BF < 1), confirming that the evolutionary lag was unique to the Xinfadi samples and not an artifact of the methodology.
Discussion and Implications
The lack of genetic divergence in the XFD genomes is most consistent with the virus being preserved in a cold environment. Given that the Xinfadi outbreak originated near a booth selling frozen seafood, the molecular evidence points directly to contaminated cold-chain imports as the source.
While the study assumes a strict molecular clock, the authors note that even under more complex "relaxed" clock models, the conclusion of evolutionary stasis remains robust. This research identifies the cold chain as a non-traditional food safety risk. For public health authorities, this necessitates a shift in surveillance strategies to include the systematic testing of imported frozen goods and their packaging to prevent future re-emergences of the virus.
Summary updated