06 August 2008

The First Search Warrant in the Ivins Case; Updated

The Department of Justice has just released documents in the case against Bruce Ivins as the anthrax killer. In this post, I want to address just the genetic and other technical issues relating to the attack material and whether it can appropriately be linked to Ivins. I will rely only on the first search warrant in the release, other relevant information may be lurking in the other documents.

The CDC examined isolates of the Bacillus anthracis extracted from the powder contained in the letters and compared it to body fluid samples collected from the five decedents. The CDC has determined that the strain of Bacillus anthracis found in each of the decedents' bodies is the same strain of anthrax found in the threat letters mentioned above. This particular strain is commonly known as the Ames strain of Bacillus anthracis.
/snip/
Determination of Origin of Anthrax Contained in 2001 Mailings

The mail attacks are believed to have occurred on two separate occasions, which are delineated by the postmark dates of the recovered letters. Letters used in the first attack were postmarked on 09/18/2001 and were focused to media establishments located in the New York area. Three weeks later, letters postmarked 10/09/2001 were sent to two U.S. Senators in Washington, DC.

In order to fully characterize the threat letters and their contents, the Task Force has conducted numerous physical (phenotypic) and genetic (genotypic) analyses. Physical comparison of the spore powders taken from the Post and Brokaw letters versus the Senator Leahy and Senator Daschle letters reveals obvious differences. The spore powders recovered fiom the Post and Brokaw letters were granular and multicolored in consistency, while the Senator Leahy and Senator Daschle letters contained fine spore powders that were uniform in color.

Microscopic examination of the evidentiary spore powders recovered from all four letters identified an elemental signature of Silicon within the spores. This Silicon signature had not been previously described for Bacillus anthracis organisms.

Much more information is needed here on the physical description of the attack material. See the immediately preceeding post for the information especially relating to "silicon". Professor Matthew Meselson of Harvard has told the FBI that silicon is present in closely related spores. The information in the search warrant is not sufficient to determine whether the Daschle and Leahy samples especially were treated with advanced materials using advanced techniques to make them more deadly. The FBI needs to release electron mircrographs of the attack material, of known highly purified spores and of spores which have been "weaponized" by the scientists at Dugway. Similarly, they need to provide raw X-ray spectrometry data of these three types of samples, as well. This will go a long way toward understanding the expertise involved in preparing the attack material.

Genetic Analysis of the Anthrax Used in the Attacks

Upon visual inspection of the Bacillus anthracis organisms used in the mail attacks, researchers have discovered numerous phenotypic variations, which distinguish it fiom the original Ames anthrax isolated in 1981, which is the earliest known sample. The phenotypic variations were identified as differences in morphology (i.e. shape, color, texture) fiom the original Ames anthrax. Changes in morphology can be attributed to individual genetic mutations within the Bacillus anthracis DNA. As a whole, the collection of all of the genetic mutations found in the anthrax used in the 2001 mailings, serve to provide a "DNA fingerprint" which can, and has been used to investigate other Ames isolates collected from laboratories possessing the Ames strain. Four individual, highly sensitive, and specific molecular assays capable of detecting four of the genetic mutations discovered in the Bacillus anthracis used in the mail attacks have been developed and validated.

Following the mailings, sixteen domestic government, commercial, and university laboratories that had virulent Ames strain Bacillus anthracis in their inventories vrior to the received Arnes strain Bacillus anthracis isolates or samples from all sixteen domestic laboratories, as well as, fiom laboratories in Canada, Sweden, and the United Kingdom. All total, the Task Force has obtained more than 1000 isolates of the Ames strain of Bacillus anthracis fiom these laboratories and archived these isolates in the FBI Bacillus anthracis Repository (hereinafter "FBIR").

The four aforementioned molecular assays have been applied to each of the more than 1000 Bacillus anthracis samples contained within the FBIR. Of the more than 1000 FB'R samples, only eight were determined to contain all four genetic mutations.

The Task Force investigation has determined that each of the eight isolates in the FBIR is directly related to a single Bacillus anthracis Ames strain spore batch, identified as RMR-1029. RMR-1029 was stored in the B3 biocontainment suite within Building 1425 of the United States Army Medical Institute of Infectious Diseases (USAMRIID), Fort Detrick, Maryland. Access to the suite is afforded only to those personnel who are approved by the USAMRIID Security, Safety, and Special Immunizations Program to have the required background check, training, and medical protection (vaccination or personal protective equipment (PPE)). Dr. Bruce Ivins has unrestricted access to the suite and has been the sole custodian of RMR-1029 since it was first grown in 1997.
The language of that first paragraph is very hard to decipher. The "phenotypic" variations they are describing here (shape, color, texture) I think must be referring to the bulk properties of the samples being analyzed and would depend far more on the culture conditions and post-culture processing than on the underlying genetics of the sample. However, the discussion does then move on to actual genetics. It appears that after original DNA sequencing of the attack material identified it as the Ames strain, the FBI then found 16 domestic and several foreign sources of live Ames strain. They then obtained over a thousand different isolates from these sources and carried out a detailed sequence analysis of all of these isolates. They report a combination of four unique mutations that are specific to the attack material. It appears that only 8 of the individual isolates in the analysis shared the four mutations with the attack material. Each of these 8 isolates appeared to have been derived from a single flask, designated RMR-1029, produced in 1997. Bruce Ivins was the sole custodian of this flask from the time it was produced.

The search warrant goes on:

Bacterial contaminant found in attack letters

Both of the anthrax spore powders recovered from the Post and Brokaw letters contain low levels of a bacterial contaminant identified as a strain of Bacillus subtilis. The Bacillus subtilis contaminant has not been detected in the anthrax spore powders recovered from the envelopes mailed to either Senator Leahy or Senator Daschle. Bacillus subtilis is a nonpathogenic bacterium found ubiquitously in the environment. However, genomic DNA sequencing of the specific isolate of Bacillus subtilis discovered within the Post and Brokaw powders reveals that it is genetically distinct from other known isolates of Bacillus subtilis. Analysis of the Bacillus subtilis from the Post and Brokaw envelopes revealed that these two isolates are identical.

Phenotypic and genotypic analyses demonstrate that the RMR-1029 does not have the Bacillus subtilis contaminant found in the evidentiary spore powders, which suggests that the anthrax used in the letter attacks was grown from the material contained in RMR-1029 and not taken directly from the flask and placed in the envelopes. Since RMR-1029 is the genetic parent to the evidentiary spore powders, and it is not known how the Bacillus subtilis contaminant came to be in the Post and Brokaw spore powders, the contaminant must have been introduced during
the production of the Post and Brokaw spores. Taken together, the postmark dates, the Silicon signature, the Bacillus subtilis contaminant, the phenotypic, and the genotypic comparisons, it can be concluded that, on at least two separate occasions, a sample of RMR-1029 was used to grow spores, dried to a powder, packaged in an envelope with a threat letter, and mailed to the victims. '
Several crucial pieces of information are missing here. First, we don't have the amount of material that is in RMR-1029 or its status in terms of processing with regard to purity or possible processing to weaponize it. In later sections of the warrant, Ivins is said to also refer to this flask interchangeably as "Dugway Ames Spores -- 1997", so it is entirely possible that the spores in RMR-1029 are treated with advanced materials. Notice also that this description does not have the previously leaked information regarding carbon-14 dating showing that the attack material was produced in the previous two years before the attack. Nevertheless, the presence of B. subtilis in the attack material is taken as evidence that RMR-1029 was the parent material for the attack material. That simply doesn't follow--if RMR-1029 contained sufficient material, the attack material could have been taken from that flask and mixed with other material, including the B. subtilis. Only with age data could one conclusively determine which sample is the parent. The presence of B. subtilis in one set of samples and its absence in another is also used to argue that RMR-1029 was sampled twice to obtain starter cultures. This also does not follow, as the two sets of attack material, if started from RMR-1029, simply could be from different stages in a purification process after culturing. Finally, the silicon content is referred to with regard to attack material but doesn't appear to be described for the RMR-1029 flask material.

The bottom line is that this search warrant does not solve many of the mysteries surrounding the nature of the material distributed through the mail. Most importantly of all, it is silent on the original source of RMR-1029. These are the critical questions that MUST be answered regarding RMR-1029:

1. Who grew that culture?
2. How large was the culture?
3. Who harvested the culture?
4. How was the material processed?
5. Where was RMR-1029 produced?
6. It it was produced at Dugway, does RMR-1029 contain all of the material produced in that batch?

Sadly, this search warrant leads to more questions than it answers with regard to the source of the anthrax attack material.

Update:

I have just noticed that the introductory "Overview" section of the warrant includes this:
At the time of the attacks, he was the custodian of a large flask of highly purified anthrax spores that possess certain genetic mutations identical to the anthrax used in the attacks;
So we do know that the material in RMR-1029 is highly purified. We still don't know if it is treated in any way to make it more dangerous. We also know that the flask is "large" but still have no idea how much material is there.

14 comments:

Anonymous said...

I have some experience in employing molecular analyses to determine epidemiology. In this case the goal is to determine whether two strains of anthrax are similar such that A is potentially the source of B.

This statement from the document sounds impressive but is welterweight: "Four individual, highly sensitive, and specific molecular assays capable of detecting four of the genetic mutations discovered in the Bacillus anthracis used in the mail attacks have been developed and validated."

This probably means each method (mismatch by hybridization or PCR) targets one mutation. It (probably) does not mean the four tests validate each other.

Furthermore, to say that they share four mutations in common by itself is a weak molecular epidemiology argument. You can always find four rare things in common between two unrelated items.

To determine that two strains have a common history should involve sequencing and summarizing all differences and similarities.

I'm not saying the better tests weren't done - just that the four mutations test (validated one thousand times in other strains) isn't strong - at least not the way it was summarized.

selise said...

at the minimum, i want to see for each of those 1000 plus samples:

1. culture, preparation and storage methods including source of inoculate, date, location, number of people with access, etc.

2. dna sequence analysis including methods used, raw data and all sequences deposited in public db (genbank or equivalent).

3. description of all other molecular assays again including methods and raw data

4. c-14 dating

5. test to determine the presence of contaminate material (especially biological contaminate). isolation and characterization (esp genetic analysis) of all bacterial contaminates. all results made public as above.

and that's just off the top of my head. where is the fucking data?

Jim White said...

Thanks for both of those comments. I agree that we are long way from having anything definitive. I really want to see a chain of custody for the generation of RMR-1029. Sounds really loose to me and it is the key to everything.

Anonymous said...

Four distinct DNA polymorphisms that identify this particular clone of the Ames strain is reasonable.

"You can always find four rare things in common between two unrelated items."

The above quote is not relevant in this case because there are large number of phenotypic characteristics that state the strain is an Ames derivative, i.e. it shares a large number of attributes with the other Ames strains. The similarities in this case are likely incredibly abundant across the panel of Ames strains tested.

selise said...

good point jim. i'll add to my list:

6. source (location, time, persons present) of each sample, method used to acquire sample. training of person law enforcement officer who took the sample (did they have a clue what they were doing?). and chain of custody for each and every sample.

Anonymous said...

Jim,
What do you or any others here in medical research think that "sole custody" mean in this context? I have the strong sense that Ivins was the equivalent of a PI -- that he ran a lab in which a small number of other, less senior scientists would have been doing most of the day-to-day bench work. But, possibly military bioweapons labs operate differently.


Holly McLachlan

Jim White said...

Holly, I think you are right that he was the Senior Scientist in charge and his entire group, and probably several other colleagues, had access.

selise and others:
I haven't waded in yet on the allegations of submission of false samples. Until we establish true chain of custody, it is really difficult to determine whether Ivins truly submitted misleading samples (which would be the strongest evidence of guilt I've seen so far) or if there is the much more nefarious possibility that the flask contents changed over time during the investigation. I realize that's real tinfoil hat territory, but it has to be considered and then there comes the question of who is behing the frame.

Anonymous said...

"At the time of the attacks, he was the custodian of a large flask of highly purified anthrax spores that possess certain genetic mutations identical to the anthrax used in the attacks;"

How would he know for sure? Was the Daschle/Leahy anthrax tested? All I've read was anthrax tested was from Mr. Stevens, victim of the first wave attack.

AS for "sole custody" that's just scare words. If Ivins had "sole ACCESS" the FBI would be screaming it night and day.

But they don't.

Anonymous said...

The anthrax from four letters and 5 deceased I believe was tested. I think the case is reasonably made in all the search warrants that he should absolutely be a suspect, which is the only thing you want out of a warrant. The warrant claims he both misled investigators and supplied them with a mislabeled sample. He also was working weekend nights in the high level biocontainment area during the attack windows, while having no reason he could give to be there, other than being out of the house, just to go to work, etc.

Anonymous said...

To quote pinko_punko:

"Four distinct DNA polymorphisms that identify this particular clone of the Ames strain is reasonable."

I agree it CAN be reasonable. But with a bacterial pathogen you will have a slow rate of mutation accumulation. Whether a strain has a particular flask as its source is not only based on four "rare" mutations they have in common, but whether or not they had other mutations that showed them to have diverged at earlier time point. If they did have other mutations that could construct another signature, then those mutations negate the inferred linkage.

They said they used their four mutation method to test one thousand other strains. That could be significant. However, the following scenario is also possible.

Finding 1. four mutations that are found in the Ivins flask and anthrax letters are not found in the other one thousand strains.

Finding 2. Six other sites of mutations are found in the letters that are not found in the Ivins flask.

Finding 3. One of the thousand strains tested (from another lab) has those six mutations in common with the anthrax letter. (Since it is one of a thousand, they are rare mutations.) A mismatch hybridization method for the four strains wouldn't pick this up AND wouldn't point to the one strain that would be a better probable source.

The way to be sure would be to substantially sequence the letter anthrax and the Ivins anthrax and see if there were other sets of four, six or whatever mutations that would be equally or more informative.

Maybe they did this. It's relatively cheap and you should have all the possible information from your key samples, anyhow.

I tried to balance this between being scientific and being accessible to the layman, hopefully I didn't do a disservice to both. Molecular epidemiology is deeper than count the mutations.

Jim White said...

quackenpuppy: I think from the other leaks over the last few days, they have actually done full sequencing of all of the 1000 or so isolates. Of course, what is really needed is for a scientist to present what they have in the analyses and what they found.

Anonymous said...

Yes, I would have sequenced a number of strains. Barring that I would have used some sort of SNP-array for hundreds of SNPs to verify that outside of the 4 determinative markers, strains supposed to have related genotypes/haplotypes did indeed have them. It is possible that the initial genotyping identified a large number of changes between a reference Ames strain and the attack strain. Following genotyping of additional strains, it may have become clear that only four markers were discriminatory between the attack strain/Detrick isolate and all other isolates. This would have been way too technical to be put into the search warrant.

I would obviously like to see the nitty gritty data. I feel like the warrants are much more thorough than the press accounts are able to portray. I find the details disseminated in the press lack the force with which they strike one when reading the actual warrants.

Anonymous said...

Jim:

I'm reasonably certain they didn't sequenced the thousand other strains. Sequencing is a much (much) more complete method of analysis, so much so that it virtually negates the statement that they used methods to detect the four mutations. A method to detect the four mutations is not even equal to sequencing four nucleotides - since you can mutate to three other forms. Sequencing the whole organism for a thousand strains (ouch) is quite a bit of work - it has 180,000 nucleotides. A thousand strains would be 180,000,000 nucleotides.

Anonymous said...

I dont have the link handy but I read somewhere that Ivins was the one that grew the RMR-1029 from samples of the original or subsequent cultures of the Ames strain. He was the sole custodian in that he was the mother of that flask but many fathers had visitation rights. I'm glad to see people with more knowledge than me looking into this because while I can kind of understand what they are claiming is fishy, the terminology throws me off a bit.