Lab Safety and the Origins of Covid-19

three clear beakers placed on tabletop

Until this day the origins of the Sars-coronavirus 2 remain unknown. The two ends of the spectrum of theories are the zoonotic theory and the laboratory theory. The first states that the virus has evolved naturally from hosts such as bats and pangolins and recently started infecting humans as well. The assumed location for this is a wet market in Wuhan. The second one assumes that the virus has escaped (or maybe even was released intentionally) from a research lab.

Any theory about the origins of the virus has to plausibly answer a couple of basic questions: How is the virus so extremely well adapted to humans (e.g. the human ACE2-receptors)? How come the virus is nevertheless not evolving rapidly at the moment? Both these things make a purely zoonotic explanation less probable, but then again: If the virus comes from a research lab, how can we not have found out about it by now?

I recently read a quite lengthy article by Jonathan Latham and Allison Wilson (L&W) that proposes a unified theory involving human passaging. If you’re up for some biology and virus evolution, it is a very interesting read. Here is the gist.

The Mojiang mine theory

In 2012 there was an incident in the Mojiang mine in China involving a disease very similar to Covid-19. Six workers got infected after they had cleaned bat faeces from the mine. All six got hospitalized with severe respiratory distress and three of the workers died. Probes were taken from the bat faeces that contained a coronavirus that was very similar to Sars-CoV2. Note that this was a virus that was adapted to bats, not humans, and that the despite its similarities, the difference to Sars-CoV2 is still substantial. Eventually probes were also taken from the workers’ lungs and thymus and send to a research lab in Wuhan hundreds of kilometers away. L&W hypothesize that some of the virus material accidentally escaped from the lab in Wuhan.

How does that answer the questions above? The key concept here is human passaging. Passaging is the process in which a virus infects a new host and quickly adapts to that host. To the virus the new host poses a hostile environment that exerts a lot of evolutionary pressure. In order to survive, the virus has to undergo quite a lot of substantial functional changes quickly. This is the phase in which viruses usually evolve very rapidly. It undergoes mutation in various sites of its genetic code – those that help the virus to adapt get passed on to the next situation. This usually results not in only one, but multiple mutations throughout the virus RNA.

The fact that we see almost no functional mutations in Sars-CoV2, even though a lot of people are infected, suggests that the passaging phase must already have happened. The virus is already very well adapted to humans and therefore does not have to change substantially anymore. This makes it unlikely according to L&H that the first passaging happened in a wet market in Wuhan in 2019. If this was the first appearance of the virus in humans, we should see much more rapid evolution.

The fact that the virus is so well adapted to the human immune system also suggests that evolution must have happened in the presence of the actual human immune system – an evolution in a human cell culture would probably not have sufficed. But how can that be? Other researchers suggested earlier that the virus would have to undergo possibly decades of evolution to become so well adapted to humans.

L&W note that a couple of things were unique about the mining incident: First, the area most severely infected was the lungs. Second, the miners were sick for weeks and months before either dying or recovering from the disease. While some parts of the workers’ lungs healed, others got reinfected in turn. This suggests a very high viral load that persisted over weeks or months. Infections with earlier coronaviruses (not the current Sars-CoV2) usually focused on the upper respiratory tract. In most cases (also including Sars-CoV2) the viral load is usually decreasing after the first couple of days. The fact that the lungs were infected is important as the pulmonary tract has about 4500 times as much surface area as the upper respiratory tract.

The idea is this: speed of evolution is largely a function of population size. The miners had a very high viral load in a lot of cells over a very long time. This, the researchers suggest, may have been an ideal condition for the virus to adapt to the human immune system. The fact that the foci of infection seemed to change over the course of time (one part of the lung healing, then another part undergoing renewed infection) is important as well: coronaviruses have the ability to evolve by recombination. It is quite possible that different virus strains have evolved differently in different parts of the lungs. They traveled, however and may have encountered each other in the same cells. Different evolution steps may then have been combined into one virus.

All of this happened in 2012. Why have we not seen any coronavirus cases before 2019? L&H explain this by pointing to the delayed construction and certification of a new Biosafety level 4 lab in Wuhan that only went into operation in 2018. They suggest that researchers in Wuhan started working with the frozen samples obtained from the Mojiang miners in 2018 and 2019. They likely may have underestimated how well the virus was already adapted to humans. So it is unlikely the virus was produced in a lab, but it may well have escaped from one.

Lab safety

But do these kind of lab accidents actually occur? Kind of really really yes. From the 2018 report of the Federal Select Agents Program (FSAP) in the US on BSAT (biological select agents and toxins, i.e. agents that can pose a threat to humans and animals):

In 2018, FSAP received 193 reports of BSAT releaseand 8 reports of a BSAT loss. By comparison, FSAP received 237 reports of releases and 9 losses in 2017, 196 reports of releases and 9 losses in 2016, and 233 reports of releases and 12 losses in 2015. As in 2015, 2016, and 2017, there were again no reports of theft of BSAT in 2018.

Or take this paper titled “Laboratory Escapes and “Self-fulfilling prophecy” Epidemics”. One of the first sections is titled “Example #1: British smallpox escapes, 1966, 1972, 1978”.

in the United Kingdom, where from 1963-1978 only 4 cases of smallpox (with no deaths) were reported from smallpox endemic areas, while during the same period at least 80 cases and 3 deaths were the result of three separate escapes of the smallpox virus from two different accredited smallpox laboratories

Note that this is after the UK ended mandatory smallpox vaccination in 1946. At the time, a substantial fraction of the UK population was therefore again vulnerable to smallpox. Example 2 talks about the re-emergence of H1N1, a virus closely related to the one that caused the 1918 Spanish flu.

It has become clear that its appearance in 1977 was almost certainly due to escape from a virology lab of a virus sample that had been frozen since ca. 1950. Only since ~ 2008 have virologists actually begun to make the suggestion of a probable laboratory release in scientific papers: “The reemergence in 1977 is unexplained and probably represents reintroduction to humans from a laboratory source”.

L&H take up this example when they talk about the 2020 Andersen et al. paper. This paper makes the case that a laboratory origin of Sars-CoV2 is very unlikely. L&H disagree and write:

Also unmentioned were instances where a lab outbreak of an experimental or engineered virus has been plausibly theorised but remains uninvestigated. For example, the most coherent explanation for the H1N1 variant ‘swine flu’ pandemic of 2009/10 that resulted in a death toll estimated by some as high as 200,000 (Duggal et al., 2016; Simonsen et al. 2013), is that a vaccine was improperly inactivated by its maker (Gibbs et al., 2009). If so, H1N1 emerged from a lab not once but twice.

Example 3, the 1995 Venezuelan Equine Encephalitis:

in 1995 a major VEE epizootic and epidemic hit Venezuela and Colombia, with a type IC virus also the cause. There were at least 10,000 human VEE cases with 11 deaths in Venezuela and an estimated 75,000 human cases in Colombia, with 3,000 neurological complications and 300 deaths. Household attack rates ran 13-57% and VEE virus was isolated from 10 stillborn or miscarried human fetuses. Full genomic studies identified the 1995 virus as identical to an 1963 isolate with no sign that this virus had been circulating and the acquiring small genetic mutations indicative of replicating in hosts for 28 years. It was another case of “frozen evolution.” […] Strong circumstantial evidence exists for an inadvertent escape in 1995, and a re-emergence in 2000 is without explanation.

Example 4, Sars laboratory escapes after the Sars epidemic in 2002/2003:

SARS has not naturally recurred, but there have been six separate “escapes” from virology labs studying it: one each in Singapore and Taiwan, and in four distinct events at the same laboratory in Beijing.

Example 5, the Foot-and-Mouth Disease (FMD) in the UK in 2007:

On August 3, 2007 an outbreak of FMD was reported on a farm in the UK, initially with at least 38 cases in cattle identified. Quarantine measures were introduced and an investigation begun, with culling of surrounding livestock. Most countries banned UK livestock and meat exports. The virus was quickly identified as a strain that had caused a 1967 outbreak in the UK, but was not currently circulating in animals anywhere. Another case of “frozen evolution.” However, this outbreak was 2.8 miles (4.6 kms) south of Pirbright, where the only two facilities in the UK that were authorized to hold FMD virus were located. […] nvestigation eventually showed that a waste-water line carrying partially treated waste water from the Merial vaccine plant to the final waste treatment plant run by IAH had gone without routine inspection or maintenance, was damaged, leaking, and had an unsealed manhole opening to the surface, so was capable of contaminating ground and surface water.

This is not good. Two things confuse me about these virus escapes from laboratories. The first is the frequency with which these escapes happen. The second is how lucky we have been in the majority of these cases. I imagine things could have gone much worse – even before Covid-19.


I don’t know whether Sars-CoV2 really escaped from a laboratory in Wuhan. I don’t believe it was engineered by humans and there is quite some evidence against that. Most of the arguments against an engineered virus rely on the fact that the virus is quite different from what simulations would have predicted to be effective (See e.g. this article). However, I find the combined theory of a virus that evolved in nature but escaped from a lab plausible.

What to make out of this? I think the take-home messages are these: First, we should at least treat this as a serious hypothesis. This means that authorities should conduct a rigorous investigation into the Wuhan research lab. Even if nothing turns up, I think it is important to take this seriously. Secondly, it is time to start caring more about biosecurity and making sure lab incidences become less likely in the future.

This problem is pressing, because it not only has the potential for large scale catastrophe, but also has a certain asymmetry inherent to it: It only needs one mistake or one careless or adventurous researcher to create a disaster. Think about all the incidences mentioned above, or think about the paper where scientists published instructions on how to assemble a smallpox-like virus from scratch.

If you are interested in working in this field (or learning more about it) I can highly recommend the 80000 hours profile on global catastrophic biological risks (there also is a link to a podcast episode). If you’re serious about this, feel welcome to contact me to discuss things and I may be able to put you in touch with some people who work on these kinds of problems.

Leave a Reply

Your email address will not be published. Required fields are marked *