December 14, 2014

VolkswagenStiftung conference on
Dual use research on microbes – biosafety, biosecurity and responsibility
December 10-12, 2014
Herrenhausen Palace, Hanover, Germany


Overview
There was little common ground concerning the GOF influenza transmission work. The flu virologists wanted to continue without any additional regulations, which was the negative buzzword for them. Other scientists pointed out the weakness of the virology and the inability to deliver on the benefits originally touted, notably pro-active vaccines.

The flu people said that these viruses were a first step in that direction and that it would take time, like so much in science. Indeed, it may take so much experimental and ferret model time to refine the studies and converge on a constellation of mutations that transforms an avian virus into something as powerful as a human pandemic virus, that it may not be worth pursuing to the end.

There is clearly a legal liability issue if ever there was an accident involving avian influenza virus GOF transmission research. For the risk experts the most likely danger comes from copying this work in less secure environments. As the methods are not too demanding, while the papers are in the public domain, comparable work can be performed. Furthermore, what is tough today will be much easier to accomplish in five years.

It is an issue of public trust. The public is primarily concerned in trusting science, rather than the scientists. Scientists have to work to build outfits that are productive and transparent. The meeting showed more clearly than ever before that there are really important issues that go way beyond the virology. Indeed they are not the remit of bench virologists who do not have the knowledge to handle them. These will have to be addressed, like it or not.


Pros and cons of the virology
The ins and outs of the influenza experiments were addressed. Dr. Kawaoka (University of Wisconsin, Madison) in particular addressed the concerns of the critics. Yet the flu virologists felt that GOF transmission research was the only way to get answers to certain questions concerning cross species transmission from birds to mammals, the ferret being the model for humans.

Nonetheless, Peter Palese (Mount Sinai Medical Centet, New York) wasn’t of the opinion that avian H5N1 influenza virus was a real threat to man, which highlights the difficulty of looking into the future of influenza infections. He spoke about “influenza viruses - facts not fear”.

Yet fear is intrinsic to influenza for three viruses have generated five human pandemics (1918 H1N1, 1957 H2N2, 1968 H3N2, 1977 H1N1 & 2009 H1N1) of varying severity taking a death toll of between 250,000 to 50 million deaths. Fear is far less associated with bunyavirus or a metapneumovirus infections, although both can kill. To quote Rob Webster as “prediction is the name of the game in the influenza field”, inevitably they are trying to predict fear.

H5N1 virus infections in man represent spillovers from animals, and they occur regularly - witness the H7N9, H6N1 and H10N8 viruses in the last couple of years.

Since March 2014 a H10N7 influenza A virus or avian origin is killing harbor seals (Phoca vitulina) in Sweden, Denmark, Germany and the Netherlands. It is the first influenza epidemic in seals recorded in Europe. The amino acid sequence at the cleavage site in the hemagglutinin molecule was PELVQGR/GLF, characteristic of low-pathogenicity avian influenza virus (LPAI). The avian H5N1 high pathogenicity avian influenza virus (HPAI) on which GOF experiments has a highly basic cleavage site.

Ron Fouchier (Erasmus Medical Center, Rotterdam) considered this new seal virus to be more dangerous than avian H5N1 primarily because it is being passaged between seals and causing death. This comment shows how priorities shift even over a short space of time - in 2010 H5N1 was the virus to beat. He considered that endless transmission of the H10N7 virus between seals was no different to passaging the virus in a GOF setting between ferrets in the lab.

In nature there is no observer bias while in the lab the experimenter introduces bias by selecting the infected animal – if the animals with low viremia (the amount of virus in the animal) are consistently chosen the end result will be a highly transmissible virus that will produce mild symptoms. By contrast if animals that have respiratory distress and high viremia are consistently selected, the experimenter will ultimately recover a highly transmissible highly, pathogenic virus. As humans are afraid of the latter, the researcher will focus on that scenario or trajectory. The question is will nature do the same?

In this context SWH (Institut Pasteur, Paris)made the comparison with dog breeding. Starting from the wolf, selection over 10,000 years ensued and then want haywire in the last few hundred years ending up with Great Danes, salukis and dachshunds, to name a few. Would nature have selected the dachshund? Given its short legs it would have fallen prey to any number of predators including their ancestor, the wolf. Lest it be said that 10,000 is a long time, Dmitri Belayev was able to select for reduced fear of humans in Siberian silver foxes over a period of 50 years. Interestingly a genetic component has been revealed (Trut et al., 2009). Observer bias is real.

RF noted that his H5N1 virus was nothing like as powerful as a human pandemic flu virus and didn’t grow as well in the ferret model. Larger amounts of virus are needed to inoculate ferrets. This contrasts with the Malta statement that his H5N1 virus “is a very dangerous virus” but in keeping with subsequent statements.

When asked why this should be and what could be done experimentally to close the gap, RF remarked that far more passaging would be necessary which would probably reveal more mutations that would ultimately transform his present GOF enhanced H5N1 virus into one of fully pandemic potential. This would take quite some time and he was more inclined to go for the major mutations rather than take the experiment to the very end.

Of course in terms of the claimed benefits this changes a lot. Since we cannot predict where these extra mutations will lie – it needs more experimentation – identifying a pro-active vaccine strain is not going to come quickly. Hence in terms of pandemic preparedness, the approach is not going to deliver fast. As drug design requires the pandemic strain this claim falls. Perhaps the data could be of use for interpreting mutations in the wild, but as the mutation constellation is incomplete, interpretation can only be partial. Once again this assumes that epistasis is not a problem.

Note also that taking the experiment to the end through repeated passaging will take it further and further on an experimental trajectory, so reducing the probability that nature will follow it among the possible trajectories open to a rapidly evolving virus.

Adel Mahmoud (Princeton University), former president of Merck Vaccines, said that flu vaccines are made according to a formula that hasn’t changed for roughly 70 years. He considered the manufacture of pre-pandemic vaccine stocks, involving manufacture of millions of doses as “wishful thinking”. In short, nothing short of THE virus or something exceptionally close to it, will work vaccine wise.

He pointed to the upcoming 2014-2015 flu season: “Increasing the risk of a severe flu season is the finding that roughly half of the H3N2 viruses analyzed are drift variants: viruses with antigenic or genetic changes that make them different from that season’s vaccine virus. This means the vaccine’s ability to protect against those viruses may be reduced, although vaccinated people may have a milder illness if they do become infected. During the 2007-2008 flu season, the predominant H3N2 virus was a drift variant yet the vaccine had an overall efficacy of 37 percent and 42 percent against H3N2 viruses.”

“The drifted H3N2 viruses were first detected in late March 2014, after World Health Organization (WHO) recommendations for the 2014-2015 Northern Hemisphere vaccine had been made in mid-February. At that time, a very small number of these viruses had been found among the thousands of specimens that had been collected and tested.”

Pre-pandemic vaccines are made in the same way and according to the same logic as seasonal human flu vaccines. This puts RF’s comment about not going to the end of his avian influenza H5N1 virus experiment in perspective – if he doesn’t go to the end then how can his virus help us?

Marc Lipstich (Harvard University) calculated the loss of life in the event of an accident. Using published data from the CDC and elsewhere, he computed that for every year in a lab working on GOF influenza transmission research the equivalent of 12,000 to 1,400,000 lives could be lost. These numbers are huge but how are these numbers to be understood.

Very simply, it supposes that several labs will be doing this work. It is impossible to predict in which lab and when a leak will occur. Given the sands of time, an accident and a lab acquired infection is inevitable (all agree that zero risk is fiction). When the accident occurs, ML’s numbers suggest that hundreds of thousands to scores of millions of people will be involved. The numbers do not mean that every year 12,000 to 1,400,000 lives will be lost. They mean that if an accident occurs in one such lab after 10 years, 10x 12,000 to 1,400,000 lives (120,000 to 14,000,000) could be lost; the risk accumulates with time and the figures add up. Obviously the more labs doing such GOF influenza transmission work the sooner the probability of a serious accident.

Again, this doesn’t indicate that any one lab will have serious accident at all. It is a collective, or overall risk. In the nuclear arena where the cat has been out of the bag for 70 years, the effort is to reduce proliferation. The same is true given the eradication of rinderpest and smallpox viruses. Efforts are made to reduce the number of labs holding and working with these viruses.

Note that even if ML is out by a factor of 100, the numbers go down to a potential of 12 to 14,000 lives lost per year of GOF research, which is still huge. For people to feel calmer, it behooves the research community as a whole to perform more risk analyses, check the assumptions and weaknesses and refine the risk estimates. Fact checking, challenging and reproducibility are essential and fast.

It is unfortunate, sad, stunning or odd that no independent risk analysis has been performed in three years of controversy. Virology shouldn’t be in this vulnerable; indeed, it is in very uncomfortable and embarrassing position. Leadership is now warranted for it has been lacking.


Biosecurity was covered by two experts: some of their comments were telling. For example, those of Paul Huntly (Risken, Singapore). “BSL3+, BSL3++, BSL4- biosafety levels are not standardized and meaningful”, “there is lots of guidance, not enough standards”, “Vast majority of accidents are caused by unsafe acts as opposed to unsafe conditions” and “the greatest risk is insiders not well controlled”.

Paul Clevestig (Stockholm International Peace Research Institute) considered that the greatest risk came from reproducing or conducting GOF flu transmission experiments in research in less stringently controlled settings compared to those performed by Drs. Kawaoka and Fouchier. The human factor is the Achilles heel.

This ties in with the immaterial nature of the genetic code. Our technology is so powerful a researcher doesn’t need to get hold of the virus itself. All you need is the published sequence: have sequence, make virus to quote David Relman (University of Stanford). Where does the scientist’s responsibility lie and/or stop?

The dilemma was illustrated by Ray Zilinskas (Monterey Institute of International Studies) who spoke about the Soviet biological weapons, circa 1972-2000, which he has followed for many years. On one slide he juxtaposed a précis of Russian bioweapons experiments enhancing bacterial and viral pathogenesis and virulence and that of GOF avian influenza research:

“1972 – Ferment’s top secret objectives were to enhance pathogens’ abilities for infection and virulence, and endow them with new capabilities to circumvent or defeat enemies’ defenses; e.g. vaccines, antibiotics, and detection techniques. Malevolent intent.

2014 – Open objectives of gain-of-function research are to increase the ability of pathogens to cause disease by enhancing their pathogenicity or increasing their transmissibility. Intent – ultimate benefit for biomedicine and public health.”

The two are remarkably similar in aims. The big difference of course lies in the intent of the two groups.

However, it shows brutally – perhaps painfully for civilian scientists – that their work can be misused by others as soon as it is published. This juxtaposition shouldn’t be taken personally and was in no way meant personally. It is just a tough reality check.

Collating a number of sentences used during the meeting leads to:
• Have sequence, make virus.
• Copying is easier than doing the original experiment.
• What is a tough experiment today will be much easier tomorrow.
• Experiments can be done by others faster than you think and in environments and countries that are beholden to their own regulations and values.
• New technologies have always been tried out for offensive purposes.

The IAP statement on biosecurity says that scientists cannot simply respond by saying that once their paper is published, they are not responsible for what others do with their findings. Yet nothing has been done by the IAP or national academies to help microbiologists come to terms with this painful dilemma. Nor was any effort made to get this message down to PIs and post docs. Microbiologists need guidance badly.

Three lawyers addressed the conference. Rudiger Wolfram (Max Planck Institute, Heidelberg) noted that freedom of scientific inquiry questions were usually provoked by the scientists. Like other freedoms, limitations arise when other freedoms are impinged upon. For example working in BSL3 or BSL4 confinement is anti-freedom, yet it allows protection of others and hence protects their freedom. He said that it was the German Government who decides the risk levels, not the scientists who provide the data allowing the decision. His simple conclusion was that is you follow the regulations you will be covered. However, the problem is that accidents generally or often arise when protocols are not fully respected.

From experience Ulrich Sieber (Max Planck Institute, Freiburg) noted that scientists accept peer review more easily than outside rules. Hence, a carrot and stick approach is necessary. He found that private/public partnerships were easier to put into place than legislation. In considering the effectiveness of self-governance he took his experience with corporate crime compliance. Studies showed that the “tone from the top” and good moral corporate standards were both effective and appreciated.

Silja Voneky (University of Freiburg) noted that in this area there were gaps in legislation that needed to be plugged. This was particularly so for low probability events of high consequence. The European Union code of human rights has an obligation to protect the individual. The gaps could be filled by codes of conduct. She recommended five levels of action that came from the German Ethics Council report – Biosecurity – freedom and responsibility of research, an excellent English version is available on the web.

• Increase awareness for biosecurity among scientists and the scientific community
• Elaboration of a biosecurity code of conduct, with a special burden of proof when undertaking such work
• Elaborate rules concerning research funding with a call for a national DURC committee
• A legal framework such that a DURC committee is consulted before undertaking such work
• International EU coordination

Volker Stollorz, a German science journalist who started life as a cell biologist, recognized the hubris in the present climate and noted that hubris is the very antithesis of prudence. He felt that society has a need to know what GOF experiments are allowed. He thought these “ruin/risk” experiments were akin to black swans and a asked what should be done. What benefits would society miss? He felt we had a serious collective action problem without any obvious solutions for the moment.

Dave Relman reminded the meeting of the moral responsibilities facing us all especially with a technology that can be outsourced more and more. What he called the packaging of recombinant DNA technology – designer genes, outsourcing of experiments to companies. Among his conclusions for the coming biological century
• Issues of risk are here to stay, and will increase in frequency, magnitude
• We must commit to ongoing conversations, relationship-building
• Scientists have social, moral obligations
• Not all “interesting” experiments should be undertaken; risks must be considered
• Mitigating the risks: raise awareness, educate, communicate, norms, guidelines, anticipate, promote flexible/agile/rapid/generic responses

The last talk was from a bioethicist. Mark Yarborough (University of California at Davis) felt the debate was about trust in scientists and about hubris. Contemporary societies have great trust in science, not the scientists. Their primary concern is the process and the benefits, not the individuals conducting the research. Trust requires transparency in the way science is performed which is why the GOF issue needs to be aired.

December 4, 2014

Revealed: 100 safety breaches at UK labs handling potentially deadly diseases
Ian Sample
The Guardian, London

A longish article detailing numerous failures in biosafety containment in UK labs in recent history, aka since 2012. They range from simple to complex failures in systems surrounding the manipulation of agents such as virulent Bascillus anthracis and Ebola virus.

Richard Ebright (Rutgers University, USA), Brian Spratt (Imperial, UK) and Tom Inglesby (U Pennsylvania, USA) provide no nonsense comments. The article is easy and worthwhile reading. 

While no system is failsafe, the simplicity of some of the "failures" reminds us that working with dangerous microbes is a more than normal risky business, and some people put their lives on the line. This is admirable. Because of this, if making microbes more dangerous is considered necessary, some sort of consensus as to the scientific merits should be forthcoming, or failing that solid reasons should be articulated that hold up to outside scrutiny. This is only fair to the researchers involved.

November 20, 2014

Openness in science is key to keeping public trust

Silence stifles progress. The scientific enterprise needs a transparent culture that actively finds and fixes problems.

Yarborough M.

Nature. 2014 Nov 20;515(7527):313. doi: 10.1038/515313a.

A must. Nothing else to be said. 

November 17, 2014

Moratorium on risky virology studies leaves work at 14 institutions in limbo
Jocelyn Kaiser
ScienceInsider

Under the Freedom of Information Act Jocelyn Kaiser obtained so-called stop orders issued by the NIAID. Dated October 21, 18 such orders affecting 14 institutions were issued covering influenza, MERS and SARS viruses. Somewhat surprisingly, a MERS coronavirus project with the aim to adapt it to mice was put on hold. The details of the project are unknown. Nonetheless it is plausible that this would involve physical inoculation of mice.

Personal opinion
This MERS virus experiment is not the GOF research most people have in mind. A small animal model of the MERS virus would be useful for testing of small molecule inhibitors and learning something of the physiopathology of the infection. Put it another way, we don’t have too many reagents for camels, which are rather large. The benefits are fairly easy to articulate.

Adapting a virus to the mouse is a goal for many researchers because there are huge numbers of markers, reagents and knock out mice that unquestionably help the scientist. Of course human and mouse genomes and physiology are different and so the mouse is “only” a model. But mouse model work invariably advances the field.

An experiment of concern is one that adapts a virus to humans and agriculturally important animals and crops. Equally increasing the virulence of an extant virus that infects humans, agriculturally important animals and crops would be of concern. Experiments of concern were first listed in the so called Fink report 2003.

November 6, 2014

Stockholm ESCAIDE 2014

Plenary session D:
Primum non nocere – Why engineer microbes to be more dangerous to humankind?

Audience survey - electronic voting via smart phones after the session

Q1: Should ‘Gain of Function’ (GOF) research be paused in the EU until clearer policies are in place for researchers?
101 yes - 43 no (70% yes)

Q2: should the public health sector be more involved in the risk-benefit analysis of GOF research?
139 yes - 7 no (95% yes)

Q3: Should the EU have a “dedicated body” to manage biosafety and biosecurity issues around dual-use ‘research of concern”?
114 yes - 29 no (80% yes)

November 5, 2014

Guest post from Prof. Mike Imperiale of the University of Michigan.

The post is on the UK Society of Biology blog ahead of their “Policy Lates” session at the Charles Darwin House, November 20, 2014.

Prof. Imperiale comments on the recent US pause in GOF virus research. He is worried by it being open ended and is concerned about collateral damage in virology.

October 23, 2014

Nature editorial – A ripe time for gaining ground

After three years of heated debate, the advocates and critics of gain-of-function research must work to agree on how best to regulate work.

Nature. 2014 Oct 23;514(7523):403. doi: 10.1038/514403a.

“And the revelations over the past few months of serious violations and accidents at some of the leading biosafety containment labs in the United States has burst the hubris that some scientists, and their institutions, have in their perceived ability to work safely with dangerous pathogens.”