Anthrax Investigation: Q&A

Q: What are the limitations of using microbial forensics for attribution purposes? Are policymakers fully aware of these limitations?
Jonathan B. Tucker, Washington, D.C.

Paul Keim: Limitations to Attribution: The conclusions drawn from any forensic evidence are very dependent upon the actual events. We frequently hear about the incredible odds against a human DNA profile ("fingerprint") matching another by random chance. This is probably the most powerful evidence that is currently admitted in criminal prosecutions. But even here, the conclusions are dependent upon the actual events. The level of certainty provided with a DNA match between a semen stain and sexual assault suspect is amazingly high and almost always leads to a conviction or exoneration. However, if there is an identical twin involved, the probabilities are not overwhelming, and you might just have to flip a coin (50:50) to decide who did it. Thus, making an accurate general statement about limitations to microbial attribution is hard to do because the specific details are critical.

However, there are some powerful specific conclusions in the Amerithrax case. These would include, for example, "The laboratory Ames strain can be precisely identified and differentiated from all other types of Bacillus anthracis, including those very close relatives isolated from the same geographic region of Texas." This is a long ways from concluding that letter spores came from USAMRIID, but it is a start and eliminates a lot of possibilities. "Exclusion" due to a lack of a match is one of the most powerful conclusions that can be drawn in DNA fingerprinting analysis.

In the Amerithrax case, the scientific evidence (morph typing) tying the letter spores to the RMR-1029 appears strong. (Disclaimer: My lab did not do this analysis, and my knowledge is secondhand.) The FBI repository of Ames cultures was extensive, and only cultures derived from RMR1029 or that culture itself have all four morphs. While this scientific conclusion excludes a very large number of possible perpetrators (including myself), it still doesn't directly attribute the crime to an individual. I understand that there was more than one person with access to RMR-1029 spore preparation. I don't know how many individuals had access, and I can only speculate that it could have been quite a few (10? 50? 100?). This limitation is well understood by the scientists involved but may or may not be understood by the public or policymakers.

Policymakers' Awareness:In general, I would say that the limitations of any technology and science are not well understood by either the public or policymakers. Scientific results almost always get simplified when communicated to a non-scientific audience, and many nuances to the conclusions get lost as well. In the end, even a jury has to decide which scientific expert is the most credible. I and other experts struggle continuously to communicate with the public and with policymakers in an effective manner. It is typical that laboratory results get communicated to a low-level policy person; who then communicates it to the next level; then to next; etc.

In the Amerithrax case, many of our scientific results were eventually communicated to President Bush. Long before he was briefed on the results, someone below him interpreted and massaged whatever report I filed with the FBI. Any U.S. president is going to be smart enough to understand the limitations, if they have the time learn them. I would guess that they never have enough time do this and that the subtleties are lost.

On the positive side, there are some very intelligent scientists and analysts in the government who do appreciate the limitations of scientific evidence. I believe that they sincerely try to communicate the shortcomings, as well as the power of a result. I get the sense that policymaking is a very adversarial endeavor and that someone will be willing to argue against any conclusion!

Q: Did any lab using electron microscopes find bentonite, or did they see silica and thought it was bentonite because they saw a brown ring?
Anonymous

Keim: This is a bit outside my expertise, but I'll relate what I have heard from the other experts: At the ASM [American Society for Microbiology] Biodefense meetings, Dr. Joseph Michael from Sandia National Laboratories reported that there were high levels of the element silicon within the spore itself. His high-resolution microscopic elemental analysis did not detect any silica or bentonite on the outside of the spores. The internal silicon signal is thought to be due to biological incorporation during the formation of the spore by the mother cell. As per Dr. Michael's talk, there was no evidence that silica or bentonite was added to the spore preps. How the silicon is incorporated inside the spore coat is not well understood but has been observed in other Bacillus spore-forming bacteria.

There is also a bonus video on the NOVA scienceNOW website involving Dr. Michael's work. [See Was It Weaponized?] Here is a good press release from Sandia National Laboratories on this topic: http://www.sandia.gov/news/resources/releases/2008/anthrax.html

Note that there has been great confusion on this topic. A very early report by Dr. Peter Jahrling (a friend of mine) with a lower-resolution technology also detect silicon in the letter spores. This silicon signal was then reported "as probably silica," which lead to speculation that bentonite was added to the letter spores to "weaponize" them. Again, this is outside my field of expertise, but so-called experts claim that bentonite can be added to spores to make them more readily dispersed. This resulted in extensive speculation that the spores were from a "state-sponsored" source and that they were of a very high "weapons grade" production. All this speculation now appears unfounded due to the over-interpretation of the early silicon signature data.

Additional Commentary: "Weaponization" is a tricky term. Most people use the term to indicate some special additive or a manufacturing process that increases the dispersion and lethality of the spores. I frequently get asked if the spores in the letters were "weaponsized." The answer has to be yes! But recognize that a brick is a weapon as soon as someone throws it at another person. That act of throwing is, in essence, "weaponizing" the brick. From what I have heard, I don't believe that the spores in the letters were "weaponized" through special milling or the additional of chemicals to increase their dispersal. I think that these are just plain purified spores that were dried and put in an envelope. Unfortunately, five people died from these "bricks." But these letters were not weapons of mass destruction. What they did, instead, was to create a major terrorism event.

Q: Did the FBI or U.S. Department of Justice consult with any genetics expert when they were asked in October 2001 whether it was okay for the [B. anthracis strains held at] Iowa State University and USDA Ames (at the strip mall) to be destroyed?
Ross Getman, Syracuse, New York

Keim: I was not consulted nor am I familiar with the actual steps that lead to the destruction of these materials. During that time period (late 2001), I was doubtlessly the most engaged microbial geneticist working with the FBI. If I wasn't consulted, then probably none others were either.

Background: This question refers to the destruction of archival B. anthracis strains held at Iowa State University during the early weeks following my lab's identification of the Ames strain from the first victim. B. anthracis cultures were fairly common in veterinary teaching colleges (ISU is a good one), and their presence at ISU is not surprising. It is a real disease, and veterinarians need to know how to diagnose and treat it. Educators would need to have examples available for teaching and reference purposes.

As we now know, the Ames strain was not from Ames, Iowa but rather from a cow that died in Texas in 1981. Even I didn't know this in October 2001 and mistakenly thought it had been a part of the U.S. bioweapons program from the 1950's. Since 1981, the Ames strain has been used for vaccine testing and, as such, is commonly used in research laboratories. Its wide use for vaccine challenges lead to the misconception that it had been used previously as a part of the weapons program. Martin Hugh-Jones and I wrote a paper in 2000 (J. Bacteriology. 182:2928-2936) where we mistakenly stated this. Martin had worked at USAMRIID and thought that this was the case when we wrote the paper. We now know that this is false.

The misnaming of the "Ames" strain was due to confusion over its original source. Dr. Greg Knudson, who was at USAMRIID at the time, has stated that it was named "Ames" because of recycled packing material used by the Texas veterinarian to ship the 1981 isolate to USAMRIID. The packaging material had originally come from Ames, Iowa, where Iowa State University and the USDA Animal Disease lab reside. This is a credible report, and we have confirmed that other B. anthracis isolates from parts of Texas are very closely related to the laboratory Ames strain (Kenefic et al, Emerging Infectious Disease 14:1494-6; Simonson et al, BMC Microbiology. 9:71doi:10.1186/1471-2180-9-71). So, we now know that the Ames strain originally came from Texas, not Iowa.

Additional Commentary: It is hard to understand why this destruction was done and why it was allowed to occur. Clearly someone in Iowa panicked at being in the media bright lights and wanted to get rid of the material. If this was really authorized by the FBI, who that authority was has not been released, to my knowledge. If the investigation had eventually lead back to Iowa, this would have been viewed as destruction of evidence and obstruction of justice. Now that would have been a public relations nightmare!

I used to work at Iowa State University, and I lived in Ames, Iowa. I still have many friends there. So that they don't get too mad at me, I would like to point out that this criticism is in hindsight, and I imagine the pressure was intense. I wasn't there in October 2001. (I had my own stress at the time.) But I hope that I would have made a different decision.

Q: The following are several of the many questions proposed at my CASE CLOSED blog, where a vibrant discussion of issues related to the anthrax investigation is active every day:

1) Did the 1,070 samples examined by the morphotype analysis accurately reflect the actual number of anthrax samples held in the U.S. in the 2001 timeframe?

2) Did the process of self-submission of anthrax samples by labs and individuals (per the FBI methodology) result in a reliable representation of the Ames samples that may or may not have existed prior to 2001?

3) Did the FBI use any type of proficiency testing for their genetic analysis?

4) Was the Bacillus subtilis contaminant found in the first batch of letters genetically identical to any forensic evidence collected from any lab? Also, was that contaminant tested against strains from Dugway Proving Grounds?
Lew Weinstein, Collioure, France

[Editor's Note: The questions above have been numbered to match the corresponding answers below.]

1) This would not be representative of all the Bacillus anthracis samples, but rather Ames strain stocks. I don't know how many B. anthracis cultures existed in the U.S. in 2001, but many more than this. The Ames strain is only one of many different identifiable strains.

2) Presumably, the 1,070 collection represents all the Ames cultures in the U.S. plus some from around the world. First, the Department of Justice subpoenaed all the U.S. anthrax labs for our inventory records with special attention to the Ames strain cultures. Next, the DOJ subpoenaed all U.S. laboratories that had the Ames strain to provide a sampling (a portion) of their Ames cultures. For example, my lab had only seven different Ames stocks, which we sampled and sent in duplicate to the FBI.

3) Definitely yes. Extensive proficiency testing was performed both prior to and during the genetic analysis at Northern Arizona University. The initial determination from the first Florida sample was done using normal research controls and standards, but soon afterwards very rigorous forensic-level protocols were implemented. There was no time for proficiency testing for the first analysis on the Robert Stevens isolate, as we had only a few hours notification. Immediately afterwards, the FBI forensic scientists began to review our procedures and help us to institute forensic-level operating protocols, including proficiency testing. Once these were in place, original analyses were repeated with the same result.

Note that there were no forensic standards for microbes and that we had to adapt the ones developed for human DNA testing to our new anthrax tests. This is a topic of great debate as to what the standards should be for a new technology.

4) Disclaimer:My lab did not work on the B. subtilis contaminant, so what I am relating is secondhand knowledge. However, it is based upon my conversations with government experts and my participation in the press conferences and the public ASM symposium. Details of the contaminant were discussed in these forums.

Background:The question refers to the non-hemolytic (B. anthracis is hemolytic) bacterial contaminant found in some of the letters. It proved to be a spore forming Bacillus subtilis bacterium. As the name implies, this is somewhat related to Bacillus anthracis and shares many biological properties. But it is not a disease causing organism, and there are many, many different types found in the environment. It would not be surprise to me to find a novel B. subtilis on my computer keyboard.

Answer:The government investigators thought that perhaps this contaminant could be used to trace back the spore preps to a particular laboratory. If the contaminant was a common laboratory strain of B. subtilis, this might have been possible. In the end, this apparently proved impossible because this contaminant was different from any known lab strains. Likewise, Dugway Proving Grounds labs were intensely investigated, and I would assume that any Bacillus strains at that facility would have been investigated. Again, no lab strains matched the contaminant using DNA analysis.

Q: Did the Bacillus subtilis contaminant found in the New York Post and NBC anthrax letters raise speculation that it could have been residue from aerosol powder equipment used by military labs to manufacture bioweapon (BW) simulants? For example, Edgewood Army labs routinely produce aerosolizable Bacillus subtilis var. niger strain bioweapons simulants. See link here: http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA483822&Location=U2&doc=GetTRDoc.pdf. Did the Bacillus subtilis found match the niger strain or any other strain associated with BW simulant production? Was this considered?
Anonymous

Keim: DNA analysis of the B. subtilis contaminant did not match any known strain of bacteria (see my previous answer).

Background:The question refers to a fairly common lab strain of B. subtilis that has been used as a non-pathogenic biological weapons simulant of B. anthracis to test detector systems in environmental releases of spores. This strain forms a very identifiable colony upon plating, which is probably why it was chosen as the BW simulant.

Everyone in the investigation is well aware of the use of this B. subtilis niger as a simulant, and many of us have this strain in our labs. Upon DNA analysis, the letter-contaminant strain was considerably different from the simulant.

Not to belabor this point, but this BW simulant bacterium has been renamed several times as scientists have learned more about. It was originaly named Bacillus subtilis var. niger, but was then deemed distinct and different from B. subtilis and was then given its own species name, Bacillus globii. Later it was renamed again to Bacillus atrophaeus. I mention this just to re-enforce the idea that this BW simulant bacterium is very distinguishable from the B. subtilis found amongst the B. anthracis spores.

[Editor's note: The following questions have been numbered to correspond with the answers below them.]

Q: 1) The mailed spores were compared to ancestral Ames (genotype62) and determined to be identical to it. In what lab and under what circumstances was the ancestral Ames stored?
2) When testing other sources to compare to the mailed spores, did you look for colony morphology differences?
3) If Iowa State University destroyed its entire collection of anthrax, how can it be eliminated as a source for flask RMR-1029?
Marcia Ann Chambers, Topeka, Indiana

Keim: 1) The ancestral Ames culture was stored at USAMRIID until it was discovered during the Amerithrax investigation. It is the oldest known archival Ames culture discovered in the investigation and dates from early 1981 (see: Ravel et al. 2009 J. Bacteriology 191:445-6.).

2) The morphs themselves were useful for identifying minor genetic differences in the letter spores, but it would have been very tedious and unreliable to depend upon visual analysis of colony morphology across the entire repository or across all the evidence. In addition, the colony morphological differences can be subtle and subjective. (A lot of credit should be extended to the skilled microbiologists who originally spotted the morphs.) So, the final repository analysis was of the genetic differences controlling the morphological differences. These were genetic tests (PCR) developed after the full genomes of each morph were determined by TIGR [The Institute for Genomic Research]. These tests are rapid, sensitive, and highly reproducible. The entire process was done in a blinded fashion where the labs used anonymous samples.

3) It is regrettable that ISU destroyed their archival B. anthracis isolates (see my commentary above), as this could have exonerated them. Our DNA analyses would have quickly determined whether they even had the Ames strain in their collection. The reason they were under such scrutiny was due to the misnaming of the Ames strain (see my discussion on this topic above).

Commentary:This is an opinion, but given the mailing location in New Jersey, it is unlikely that an ISU employee could have driven that far to mail the letters without being detected. It is also unlikely that ISU even had the Ames strain, let alone RMR1029. They did not have a vaccine development program, and there were no notable anthrax research teams there at that time. After all the unwarranted media attention, I'm guessing that the government investigated activities at ISU with great diligence and ruled out potential suspects.

Q: Dr. Keim,
In the segment on microbial forensics June 30th, the morphs were eventually found, which distinguished eight of the original 99 lab samples as being identical to the Ames strain RMR-10-20-9. When the initial comparison of the >5 M nt were done, comparing the anthrax in the dead journalist and the original cow anthrax, why was the original verdict that they were exact matches? Why were the morphs not detected in early 2002? Had the one scientist not seen a slight color variation, would the morphs never have been found? Were all the morphs found in the same petri dish together? Using today's technology, would we be able to find the morphs via the sequencing matches?

Also, can you tell me what minimum undergraduate and/or graduate courses would have to be completed in order to be considered as a graduate student candidate in microbial forensics? I believe yours is the closest university that offers such a program in the Southwest U.S. Thanks.
Diana, Albuquerque, New Mexico

Keim: [Addressing the first paragraph:] The morphs were observed very early in the investigative process. I wasn't there when it happened but probably in 2001 and 2002, the morphs had been observed. The observation probably occurred when a microbiologist was plating for single spore colonies on a blood agar petri dish. Hundreds of colonies could be grown in this fashion, side by side. Subtle difference in a few colony morphology were spotted at this time. We are told that they were in low frequency (1%?), so a fairly large number of spore must have been platted (grown) in order to see the few that were different.

With today's genomic technology, it might be able to identify minor component genetic differences without first spotting the morphological mutants. The morphs themselves are of little value other than pointing to genomics differences that could be used for characterizing the samples. Today we could simply sequence a large number of colonies to find differences.

[Addressing the second paragraph:] Microbial forensics is a new scientific field but closely related to the field of epidemiology. In particular, forensic analysis similar to molecular epidemiology analysis was important in the Amerithrax investigation. A graduate program that covers infectious diseases, microbiology, molecular biology, genetics, genomics, epidemiology, evolution, statistics, and public health would be excellent training for a budding microbial forensic scientist. There is a great need for molecular epidemiologists to understand and solve public health problems. I am hoping that there will not be a great need for microbial forensics in the future. If so, some of the talent pool will come from epidemiology.

Q: Something did not make sense in the documentary. You narrowed the source to a particular bottle, but also stated that about 10 labs had tapped it. Why then was Bruce Ivins, the keeper of the master bottle, the only suspect? Why not one of those other labs or Ivins' superior or coworker?

Also, the NOVA piece did not speculate about motives for the attacks. Surely that would also be a way to narrow down culprits. As Seneca put it: Cui prodest scelus, is fecit (The one who derives advantage from the crime is the one most likely to have committed it).
Roedy Green, Victoria, British Columbia, Canada

Keim: The scientific evidence from the morphs only narrows the search to RMR1029 and cultures derived from it. In the 1,070 FBI repository samples, there were eight samples that contained all four morph signatures, most of which were at USAMRIID. It is my understanding that there was only one other lab, besides USAMRIID. This is secondhand knowledge, on my part.

I have no special information concerning the attribution to a particular individual and can't answer the second two questions.

I have no special information concerning the motives of any individual and can't answer the final question.