The Sunshine Project
Biosafety Bites #21 (v.2)
Updated (26 January 2007)
Originally published on 11 January 2007, Biosafety Bites #21 describes an incident involving a genetically engineered cross of H5N1 ("bird flu") and H3N2 influenza that occurred in a BSL-3 lab at the University of Texas at Austin (UT) on 12 April 2006. It is based upon UT records, specifically, UT's incident report of 13 April 2006, biosafety committee minutes, e-mails, and letters acquired under the Texas Public Information Act.
After publication of Biosafety Bites #21, UT was contacted by several journalists. UT provided information to the journalists that contradicts what is stated in UT's records. UT's contradictions prompted further uncertainty about the facts of the incident and how UT managed it.
Seeking to clarify the public record, on 19 January 2007, the Sunshine Project submitted ten questions to UT's Vice President for Research, Juan Sanchez. On 25 January, Sanchez replied with answers to the questions and two additional items - a new version of UT's accident report, and document titled "Message for Lab Incident" (talking points used by UT's public relations department).
The Sunshine Project believes that UT's reply remains deficient in many respects. What actually happened? We are unconvinced that there was no public health threat. For example, UT now claims that none of its staff knew the identity of the virus that was being researched until 5 days after the incident. This does not sound plausible. Has UT handled this accident and requests for information about it well? We believe the answer is an emphatic "no". Rather than being striaghtforward, the University has demonstrated a disheartening pattern of misleading statements and intellectual dishonesty. (For example, about the nature of its work with allegedly "non-infectious parts" H5N1.)
Is UT trying to cover up a more serious accident than it has admitted to? We believe that there is a strong possibility that this is true. Readers are invited to judge for themselves. Below, we have posted the relevant documents and the Sunshine Project's rebuttal to the information provided by UT's Juan Sanchez in late January 2007.
A final note. The original Biosafety Bites made no reference to the names of individuals involved in the incident. The Sunshine Project sought to keep discussion at institutional and policy levels - and not to prompt a discussion involving individuals. UT, however, responded to the Biosafety Bites by refuting information contained in its own records. This placed the Sunshine Project in a position where it was necessary for us to post our sources. These UT sources contain the names of people involved. This is not how we would have done things; but UT's contradictory positions with respect to its own records leaves us little choice.
13 April 2006
UT BSL-3 Laboratory Incident Report (PDF)
UT was instructed to release this report (and the e-mails below) by the Texas Attorney General.13 April 2006
UT Health and Safety Office E-Mails on the Incident Report (PDF)
"Other than the lessons learned, I think this report is complete."17 April 2006
Minutes of the UT Institutional Biosafety Committee Meeting (PDF)
Containing an account of the incident inconsistent with the accident report.26 September 2006
Letter from UT Vice President's Office to the Sunshine Project (PDF)
Stating that no state or federal reports were filed.3 October 2006
UT Legal Brief to the Texas Attorney General (PDF)
"... the University asserts that [Texas statutes] make information responsive to the request confidential by law."8 December 2006
Letter from UT Vice President's Office to the Sunshine Project (PDF)
Releasing accident report per Attorney General's instruction, providing no indication that UT disputes facts contained therein.11 January 2007
Sunshine Project Biosafety Bites #21 (html)
Providing our interpretation of the UT records.19 / 25 / 26 January 2007
Ten Questions for UT, UT's reply, and the Sunshine Project's rebuttal (html)
The Sunshine Project
Biosafety Bites #21 (v.2) - 11 January 2007
The Bird Flu Lab Accident that Officially Didn't Happen, or
How the University of Texas at Austin Could Have Caused the Next Influenza Pandemic, but Everybody Lived to Cover It Up
Don't ask the National Institutes of Health (NIH) about the genetically engineered influenza pandemic that might have started in Austin, Texas in April 2006. That's because until NIH reads this Biosafety Bites, they almost certainly haven't heard anything about it. And that shows yet again that the US biotechnology and laboratory safety oversight system is a dangerous failure.
NIH's Office of Biotechnology Activities (OBA) doesn't enforce biosafety rules, so the University of Texas (UT) didn't report the unsettling Bird Flu accident. UT must have reasoned: Why draw attention to a lab accident when there's no cost for burying such incidents? It surely wouldn't be the first time such an event has been swept under the rug.
BSL-3 in the Heart of Texas
According UT records obtained by the Sunshine Project (PDF link), the accident happened on a Wednesday afternoon, 12 April 2006. A postdoc was working in the Molecular Biology Building ("MBB") on the University of Texas campus in Austin, just a couple minutes' walk away from tightly packed dormitories, the kind of place where a virulent new influenza strain might eagerly take hold. A little over a kilometer south is the Texas Capitol and a warren of state office buildings teeming with public employees.Centrifuge Accident Aerosolizes Genetically Engineered Influenza
The postdoc was working alone in a beefed-up BSL-3 laboratory wearing a full lab suit. A respirator system provided oxygen through an air hose. The high-tech safety measures were in place because the viruses in the lab were not your average flu. They were something much more dangerous. They were genetically engineered influenza strains that mixed and matched genes of the common human H3N2 influenza and those of deadly H5N1 "Bird Flu". The kind of unpredictable reassorted flu strain that public health officials fear could cause the next human pandemic.In the BSL-3 lab, a quantity of the engineered influenza was ready for work. It had been grown mixed with cells. The experiments required purified virus. So, a little after 2:00PM, the researcher transferred a quantity of the virus mixture into a tube. The tube was capped and placed in a centrifuge on a lab bench. The centrifuge would separate out the virus through spinning - centrifugal force.
But the tube was of the wrong type for the centrifuge. There were two almost identical centrifuges in the lab, and their non-interchangeable parts had become mixed up.
The postdoc pushed a button and the centrifuge began to spin. Because the tube was the wrong type, its cap didn't fit correctly. It cracked. The centrifuge lost balance. Turning the machine off and then opening it, the postdoc observed that the level of virus fluid in the tube had gone down and that its exterior had become wet, both indicators of a leak. This was a serious problem because as the machine spun around, the leaked virus had become aerosolized, at least within the centrifuge.
The Inevitable Human Error
By now the cracked cap problem had been compounded by human error, an ever-present factor in lab work. Rather than waiting for the aerosolized flu to settle, the centrifuge had been immediately opened. In an invisible puff of air, virus particles wafted out of the machine. Now, the virus was floating around the whole lab, stirred by air movements, then slowing settling on exposed surfaces or being sucked out the exhaust which, hopefully, had effective HEPA filtration (the UT documents are silent on this item).It was something like a Bird Flu victim walking into the room and coughing all around, spreading virus where he went. Except this mixed up lab creation of H5N1 virus was possibly more efficient at infecting humans than natural "Bird Flu" because of its H3N2 human influenza parts.
The researcher sprayed Lysol and wiped up surfaces in the work area, exited the lab, took a shower, and put on new clothes. Within hours, the postdoc was taking Tamiflu, in the hope that it would stop the virus if the researcher had been infected. For several uncomfortable days, the University of Texas staff waited to see if the researcher developed symptoms. None are reported to have appeared.
The University of Texas at Austin had dodged a bullet. It took longer for a UT biosafety team to straighten out the lab and reopen it. Under any of a variety of plausible scenarios, the accident might resulted in disaster. For example, if the cap leaked but didn't crack, without the postdoc noticing, thereby multiplying the danger to include everyone working in the lab over a longer time.
UT's Bird Flu Hybrid and Deceptive Records
Reading UT's records (view the PDF accident report here), it is clear that the University was thinking in terms of public relations from practically the moment that the accident occurred. UT records unscientifically discuss (downplay) the risks and neglect to precisely describe the flu strain. For example, they state that the virus should be considered like far less dangerous H3N2 despite it being a hybrid with "some genes from H5N1". This is deceptive, because the bug that causes flu is composed of only 8 short pieces of RNA that collectively encode just 11 proteins.Assuming "some genes from H5N1" means at least three RNA pieces or more, or the RNA to encode three proteins, UT's hybrid Bird Flu virus would be about 25% H5N1 (somewhere between 3/11ths and 3/8ths), and potentially much more if the "some genes" were larger ones. That's certainly enough H5N1 genetic material to create an unpredictable and potentially extremely dangerous (pandemic) reassortant. Tiny differences in genes can make huge differences in the bug. Nobody knows for sure how dangerous UT's flu was because, by good fortune, this story doesn't end in human infection.
UT's report also deceptively states "CDC recommends BSL2 practices for H3N2, but it was decided that BSL3 would be prudent for use with this agent," as if UT was acting with an abundance of caution. But UT was was working with a potentially pandemic combination of H5N1 and H3N2. And well before April 2006, there had been scientific discussion and government recommendations made about the need for BSL-3 or higher containment for flu viruses like UT's. Thus, contrary to the implication of its PR-wise assertions, UT was not taking any major steps above and beyond the basic measures that should have been used for such a virus.
Echoes of 2005's Flu Accident
It must have weighed heavily on the minds of University of Texas public relations officials (who were called less than 2 hours after the accident) that one year before, on 12 April 2005, global headlines were dominated by the story of Meridian Biosciences Inc., which sent 3,700 samples of potentially dangerous noncontemporary H2N2 flu to labs in the US and across the world. If the UT accident became public at that time, its occurrence on the anniversary of the Meridian story might have cast an extra bright and unflattering light on the University of Texas, potentially unsettling the Molecular Biology Building's many neighbors, many of whom would be unhappy to learn that they came too close for comfort to being ground zero of a deadly flu pandemic.Need for Federal Reporting
Although it would serve public health and accountability ends, perhaps it is presently optimistic to expect a university to quickly issue bad news about itself, especially when that bad news evokes images of it authoring a public health disaster. But it must be expected that such accidents definitely will be reported to the federal officials that oversee lab safety so that, at least, other labs can learn from the mistake and, for example, not put two identical centrifuges whose parts are NOT interchangeable in the same lab. And so that federal safety officials and funders could examine the accident and impose penalties if institutional safety deficiencies are identified.Accident, Revised Out of Existence
But it does not appear that anybody outside UT found out about the incident until the Sunshine Project requested the accident report. UT fought to keep it under wraps. While the Texas Attorney General's office was weighing a UT petition to keep the accident details secret, somebody got cold feet. A UT official left two messages on the Sunshine Project answering machine offering to explain what happened, if the Public Information Act request was withdrawn. (We did not respond.)The Public Information Act request revealed that UT never finalized its accident report and it did not inform NIH. Instead, it made the accident disappear.
How? On the morning after, officials interviewed the postdoc. Remarkably, they recorded that the postdoc's account of the accident had dramatically changed overnight. UT's Environmental Health and Safety Office writes "The researcher thought that the volume of the tube had changed, but was not 100% sure of the original volume." The liquid on the exterior of the tube? It "may have been from condensation". The lid? It, at least, was still broken.
The accident was miraculously converted into a figment of the postdoc's imagination. Pondering the possibility of being at the center of an embarrassing incident that might impair funding and anger UT leaders, was there pressure to change the story? The postdoc knows for certain; but in the absence of any enforced reporting requirements, there were precious few incentives to move forward with accident reporting. Or perhaps UT management insisted that nothing happened unless the Tamiflu-taking postdoc affirmed absolute certainty of details remembered while in the midst of scrambling to contain a potentially life-threatening accident?
Certainly, UT management seized upon the (reported) "not 100% sure" statement. On that basis UT decided that an accident had not occurred. The following gem of illogic (read carefully) provides the University's reasoning that the accident didn't happen: "There is the possibility that there was no leak and therefore no contamination occurred."
The following Monday (17 April), UT's Institutional Biosafety Committee (IBC) held a previously scheduled meeting. The incident was briefly discussed. In the IBC minutes (PDF link), a new version of events appears, one that omits several critical details from the accident report. According to the IBC account, the postdoc's concern was said to have been that the tube (not cap) had cracked, but that thankfully, it hadn't. It was a mistaken impression by the young researcher. The tube was fine. And "the liquid on the tube"? It was "probably condensation". The broken cap isn't mentioned. Nor is the prematurely opened centrifuge. Nor is the decrease in the volume of the virus in the tube.
Condensation? According to the accident report, the "condensation" was observed not long after the tube was filled and almost immediately after it had been spinning at several hundred, perhaps several thousand, revolutions per minute. If it was condensation and not virus culture, then UT seems to have set a world laboratory record for the fastest-forming and most remarkably adhesive water condensation ever seen.
But as far as UT was concerned, the case was closed. No authorities were told. Officially, no accident took place, although despite the fact that nothing officially happened, UT curiously proceeded to decontaminate the entire lab "as if the contamination had occurred." The accident report remained labeled "draft" and was not finalized.
And there the story would have ended, before this Biosafety Bites.
10 Questions for UT, UT's Reply, and the Sunshine Project's Rebuttal
The following questions were sent to UT on 19 January, after it disputed details of Biosafety Bites #21. UT's Vice President for Research, Juan Sanchez, replied on 25 January.
Question 1: Recently, UT has stated that the virus only contained one gene from H5N1 and has specified that this gene was a non-structural gene. Yet the UT accident report refers to "some genes" from H5N1, and the minutes of the UT Institutional Biosafety Committee specifically state that "The virus had H5N1 structural proteins included". Please explain these discrepancies and provide the genetic composition of the virus involved in the accident.
UT Reply: The Institutional Biosafety Committee (IBC) meeting occurred shortly following the incident. A clarification as to the exact contents of the vial was provided after the IBC meeting. The minutes will be amended to reflect the clarification. The virus was composed of H3N2 A/Udorn/72 with the NS gene substituted with H5N1 A/Hong Kong/483/97.
The incident report submitted to the Sunshine Project, in response to the request under the Freedom of Information (FOI) act, was clearly marked “DRAFT” in all its pages. The report was drafted during and shortly after the incident. At the time the report was drafted, it had not been determined which H5N1 (non-HA, non-NA) gene had been substituted. It was later determined that the single H5N1 gene that had been inserted was NS.Sunshine Rebuttal: UT's response describes a remarkable, perhap unbelievable, state of ignorance about the virus for many days after the incident. The timeline:
12 April (PM): Incident occrus, researcher interviewed.
13 April (AM): Researcher and PI interviewed, lab examined
13 April (PM): On the basis of those interviews and lab inspection, the UT accident report states that the virus contained "some genes" from H5N1.
17 April: The minutes of UT biosafety committee state that the virus contained H5N1 structural proteins (plural). These minutes were approved at the next IBC meetingSometime (unknown) after 17 April: UT says it discovers the genetic composition of the virus, and the previous descriptions of it were wrong. According to UT's January 2007 statements, its initial report and IBC minutes were both wrong. How is it possible that nobody at UT knew the genetic composition of the virus until more than five days after the incident?
With respect to the maturity of the investigation information contained in the 13 April incident report, UT staff e-mails released under the Texas Public Information Act state: "Other than the lessons learned, I think this report is complete."
The Attorney General of Texas was provided all UT records on this incident, including records that UT has not released to the public. With all of UT's information at his disposal, the Attorney General determined that the draft was a factual account of the incident and, accordingly, denied UT's request to withhold it as a preliminary "opinion" record.
Question 2: If the University is confident that no leak occurred, why was the researcher placed on Tamiflu and continued to take Tamiflu for a week, and;
UT Reply: The researcher was immediately put on Tamiflu as part of the BSL3’s Standard Operating Procedures (SOP) for spill incidents. It was determined that a leak had not occurred the next day. The researcher was suggested to continue the medication as a precautionary measure.
Sunshine Rebuttal: In point of fact, the accident report of the next day (13 April), does not state that it was determined that a leak had not occurred. It states that there is "the possibility" that a leak did not occur. With such viruses, a SOP for spill s that calls for antivirals makes sense, however, keeping the researcher on Tamiflu after it was determined that a leak did not occur does not. If there was no leak, then there was no reason (and it would be medically inappropriate) to continue to take the drug, whose side effects include nausea, vomiting, diarrhea, bronchitis, abdominal pain, headache and dizziness.
Advising the researcher to continue the take the medication was the logical course of action if the University believed that there was significant possibility that the researcher was exposed.
Question 3: Why did the University decontaminate the entire lab "as if the contamination had occurred"?
UT Reply: Decontamination of the lab is SOP when a researcher thinks there has been a spill. It was later determined that a spill did not occur.
Sunshine Rebuttal: University safety staff performed the full lab decontamination beginning at 11:30AM on 13 April, after interviewing the principal investigator and the researcher (twice). If the University had determined that no leak occurred on the basis of those interviews, then there was no reason to proceed with the decontamination.
Question 4: The University states that it is confident that prior infection with H3N2 influenza would immunize against the virus used in the lab. Was the researcher vaccinated against the H3N2 strain before initiating the work? If not, why not?
UT Reply: All researchers (including this one) are required to have the current year’s influenza vaccine, which contains H3N2 antibodies and provides partial immunity to the laboratory strain.
Sunshine Rebuttal: The reply does not state that the requirement was complied with. If H3N2 antibodies from contemporary circulating flu strains only confer partial immunity to the laboratory strain, why does the University continue to trivialize the distinctions between H3N2 and the virus involved in this incident?
Question 5: The UT accident report states "University researchers DO NOT work with H5N1. Our researchers work with non-contagious elements of that virus." Why was the University handling these "non-contagious elements" of H5N1 at BSL-3?
UT Reply: CDC recommends BSL2 practices for H3N2, but it was decided that BSL3 would be prudent for use with this transgenic agent.
Sunshine Rebuttal: The reply does not address the question. The agent in use was not H3N2. It was a combination of H3N2 and an H5N1 strain isolated in 1997 in Hong Kong following an outbreak that killed 6 of 18 persons infected. A year before the UT incident (April 2005), a joint CDC-NIH interim recommendation urged a "cautious approach utilizing elevated biocontainment levels and practices" for such viruses. To describe UT's work with H5N1 to be with "non-contagious elements" of H5N1 is very misleading.
Question 6: Neither the accident report, nor the IBC minutes, nor the University's recent statements to reporters are factually consistent. Why is the University failing to produce a consistent set of facts and description of this incident?
UT Reply: The accident report, IBC minutes, and the university’s recent statements to reporters are factually consistent with the exception of plural versus singular references to the word “gene.” The attached final report and prepared media messages reference a single H5N1 gene not some H5N1 genes.
Sunshine Rebuttal: If these items are factually consistent, then UT claims that the H5N1 NS gene encodes structural proteins (it does not), that "the possibility" that an accident did not occur demonstrates that an accident did not occur (an obvious failure to reason), and that H3N2/H5N1 reassortants are non-infectious, among other falsehoods that UT continues to propagate.
Question 7: With the sole exception of the IBC minutes, which must be released under federal guidelines, why did the University petition the Attorney General for permission to keep every single one of its records about this incident a secret?
UT Reply: It is the university’s procedure to seek guidance from the Attorney General when FOI requests for research-specific information are received. Proprietary information, which is the intellectual property of the researcher, was redacted from the disclosed documents as a result of the AG’s decision.
Sunshine Rebuttal: In its brief filed with the Attorney General, UT stridently sought to withhold all records. UT is wrong again on a point of fact. The Attorney General did not permit redaction of any of the disclosed documents for reasons of proprietary information. The sole redaction permitted by the Attorney General in the disclosed documents was of the "lessons learned" section of the accident report which, at the time of his ruling, was incomplete. This redaction was under Section 552.111 of the Texas Government Code, and pertains to opinions offered for policymaking (the draft "lessons learned"), not proprietary information.
Question 8: Why did the University not report this incident to the National Institutes of Health, whose Guidelines require (Section IV-B-2-b-(7)) that accidents be reported?
UT Reply: No significant research-related accident or illness had occurred, so none was reported to NIH.
Sunshine Rebuttal: It is clear that it cannot be concluded that no significant accident occurred. Moreover, UT's actions (human errors, tamiflu, decontamination, prolonged confusion about virus identify, secrecy, misstatements of fact) strongly suggest that one did. If UT desires to act prudently, it should report the accident.
Question 9: What specific facts contained in the University of Texas accident report released to the Sunshine Project does the University of Texas now claim are incorrect? Please specifically and clearly enumerate them.
UT Reply: The incident report was a real-time draft provided to the Sunshine Project. The final report clarifies that there was one H5N1 gene in the vial not some H5N1 genes.
Sunshine Rebuttal: As previously stated, the 13 April accident report was described as "complete" by UT staff, and was ruled to be a factual report by the Attorney General. It is a very strange procedure for the University to have completed a final draft in April 2006 yet not finalized it until much later. UT claims that the accident report was made final on 10 October 2006, although the file provided to the Sunshine Project by UT's Juan Sanchez contains Microsoft Word metadata that indicates that it was edited as late as 25 January 2007.
Question 10: If the University's accident report is seriously flawed, why did the University release the report with no indication that it contains errors?
UT Reply: The university provided prepared documents as of the date of the FOI request. The incident report was a real-time draft provided to the Sunshine Project. As stated previously, the final report is attached for your information.
Sunshine Rebuttal: This is amply discussed in our reply to prior items. The draft was complete except for "lessons learned", was ruled factual by the Attorney General, and was completed after investigation of the lab and interviews with the researcher and principal investigator.