I have now had a chance to read the PNAS paper by Maines et al. and it is surprising in two respects. The first is it isn't that interesting. The second is related to the first. Why did they bother to hold a press conference about it? Even more, why did the press conference focus on the reassortment question when that didn't establish much. Anyway, that's how I read it. Here are my reasons.
The most important part of this paper is methodological, testing a ferret infection model for transmissibility. Ferrets have been used as a reasonably good biological model for infectivity and virulence in humans so their use in this experiment makes sense. Transmissibility was modeled by putting infected ferrets in a cage separated from uninfected ferrets by a perforated wall. This prevented direct contact but allowed air to circulate between the two spaces. If infection does pass this boundary, it is reasonable to say the infection has passed from one animal to the other via the air. If it doesn't, this still leaves open the possibility of an airborne transmission not detectable in this model. As the authors point out, this establishes nothing about the physical size of the airborne virii that moved from one portion of the cage to the other, an important question not tested here.
The experimental results showed that a human H3N2 subtype strain known to be infectious in humans could be transmitted from infected ferrets to ferrets on the other side of the perforated wall. Thus transmissible human viruses were transmissible in the ferret model. One of the H5N1 strains from the 1997 Hong Kong outbreak was transmitted to other ferrets in this fashion (by air) also, using seroconversion as evidence of infection. When ferrets were put in the same compartment with ferrets infected with this 1997 virus, the researchers were able to find seroconversion in two animals and viral shedding in one. Thus the H5N1 from Hong Kong could be transmitted horizontally but wasn't as efficient in doing so as the human H3N2 subtype. Since current H5N1s are genetically distinct from this original 1997 virus, three other H5N1 strains were tested in this way. An Indonesian isolate and two isolates from Vietnam showed no transmission by any measure. thus the ferret model predicts H5N1 will not be very transmissible in humans and it isn't (so far). The Indonesian isolate was from May 2005. It would be interesting to do this experiment with the isolates from the recent Karo cluster in Indonesia (May 2006), because this was an instance where human to human transmission is fairly certain. If the two viruses from Indonesia a year apart showed different results in the ferret transmissibility model this would be evidence both that the model is a sensitive predictor of transmissibility and that something in the virus had changed in the space of a year. The hunt would then be on to see what it was.
The data in this paper also show the virus in the infected ferrets was being shed in the upper respiratory tract. The whole question of where in the respiratory tract of ferrets and humans there are cells with human or avian receptors is still unclear. This paper repeats the statement that ferrets and humans are similar in having predominantly α2,6 receptors in the upper respiratory tract, but this conflicts with data in a paper by Matrosovich et al. (2004) and a recent one by Ibricevic et al. (2006) showing α2,3 receptors in human ciliated cells, which are in the upper respiratory tract. These data on ferrets suggest the same is true for this animal. The significance is that failure to transmit efficiently probably does not rest with the explanation that only cells deep in the human lung are infected. The mystery of transmissibility lies elsewhere.
This is also indicated by the other main results of this paper involving reassortment, the part that has been covered in the press but in our view of lesser significance. There are many ways that combinations of the surface proteins and internal proteins can be produced by shuffling. If we look only at the H5N1 half and ask how many ways combinations of partially avian and partially human internal proteins can be obained by reassortment of the internal proteins, the answer would seem to be 63. There are six segments left after we fix H5 and N1 and 64 ways to get combinations of two viruses in those six pieces, one of which is the original six of the H5N1, leaving 63 other combinations. The paper doesn't really tell us how many combinations were tried. Data in Table 2 give results for just four of them and a mention is made of another one, suggesting other combinations were also tried and not reported. Whatever the number, it is likely to be only a small fraction of the possibilities for just these two strains of these two subtypes. If we do this for other strains of H3N2 or H5N1 and other human subtypes (like H2N2 or H1N1) we might come up with different answers.
Science can't do everything at once and we don't fault the investigators for that. But it seems quite a lot was made out of a restricted question and that produced tentative answers about reassortment. It is hard to see how this paper establishes much about reassortment in general and certainly says nothing about other sources of genetic variation within the segments. An interesting side result involving passaging one of the reassortants five times through ferrets. It showed essentially no adaptation (in fact the degree of genomic stability was very surprising).
Yet we don't know if we are just one point mutation away from an important change to efficient transmissibility or ten or more away. One thing the reassortment did establish is that ability to replicate and transmissibility arenot the same. Some of the reassortant viruses replicated easily but didn't transmit easily, again showing that some of the easy answers about why the virus is easily transmissible (e.g., it infects cells deep in the lung) are just speculation and probably wrong.
We are still guessing about this virus.
- Log in to post comments
With regard to using the May, 2005 isolate from Indonesia, the data on H2H is pretty clear. The isolate is from the father of two daughters who died (and is the official first human case in Indonesia). One daughter was the index case and she had very high titers of antibodies agains H5N1 (and I think recently she became an official confirmed case). All three family members died and disease onset dates indicted H2H. Moreover, the Indonesian isolate represents most of the human cases in Indonesia.
The Karo cluster is somewhat of an anomoly and most of the data in the paper was genrated several years ago. Adding an isolate from Karo would take some time and frequently the target H5N1 is somewhat limited by how well the H5N1 replicates in culture.
In addition, the 2005 H5N1 is the target for one of the pandemic vaccines under development.
If the current clusters in Karo end up being positive and the H5N1 is similar to the H5N1 from the earlier Karo cluster, I suspect similar tests will be run with a representative of that sub-strain, which only has limit cross reactivity with the 2005 isolate.
Revere. A wonderful piece of writing: Thanks.
I am still in awe at the wide range of irresponsible headlines the paper spawned...
IE. http://wpherald.com/articles/638/1/CDC-readjusts-bird-flu-pandemic-like…
You / Revere wrote: "If infection does pass this boundary, it is reasonable to say the infection has passed from one animal to the other via the air."
I don't think so.
Perhaps semantics but, semantics is part of communication and, communication is everything ....
The Infection Control community DEFINES sneezing / coughing modes of transmission, occurring at separation distances of less than 1 m, as being "contact", and NOT "airborne" -- e.g. see http://www.cpso.on.ca/Publications/infectioncontrolv2.pdf .
Regardless of anyone's "take" on this, I think that it is important for all to understand the significant limitation of the study's mechanical experimental design in permitting / assessing airborne transmission.
You assumed: "... allowed air to circulate between the two spaces." Au contraire!
The experimental set-up (not described in detail) had the animals "within a Duo-Flo Bioclean ...".
In this, there is a laminar flow of air in a VERTICAL direction i.e. minimizing / preventing the lateral / horizontal movement of air (e.g. between adjacent cages).
How effectively this works depends on details such as cage construction and airflow obstruction by bedding, etc.
But, the question(s) that may be of the essence:
Was it (exclusively) the SNEEZING of the ferrets (and, consequently the projectile dispersion of mucus gobs) through the perforated grilles that caused sero-conversion?
And, contagious particles of respirable dimension were effectively controlled by the system -- as it was designed -- i.e. carried vertically away from those who might otherwise have become infected by inhalation?
And, this paper therefore has nothing to do with what we would reasonably consider airborne / respiratory (as in, "breathe in, breathe out, repeat as necessary") contagion?
Dr. Niman or Revere - can you write a bit more on the difference between reassortment, recombination and mutation for the non-scientists reading all this?
ugis: Thanks for the comment. I saw your concerns about air flow in the other post. I think there just isn't enough information in the paper to tell if that's an issue here or not. If the air flow were truly laminar it is hard to see how there could have been any transmission between the compartments which is why I wasn't so bothered by the possibility. But I don't think the paper was well reviewed as there are a number of clarifications like that that should have been caught by peer reviewers (I speak as a journal editor, here).
K. In the previous post on the paper (a day or so ago) Henry responded to this question in the comments. I think his explanation is good. If you haven't seen it, take a look.
Re: "If the air flow were truly laminar it is hard to see how there could have been any transmission between the compartments ..."
Actually, it has less to do with laminarity and more with velocity.
Regardless, sneezing propels large particles across airstreams / gaps that could not be directly traversed by respirable particles.
Please see
http://jacobsmedia.typepad.com/photos/uncategorized/sneeze.jpg
and/or
http://www.rit.edu/~andpph/photofile-c/sneeze-k-17.jpg
Sorry -- couldn't locate any pics of ferret sneeze projection, but they are seemingly rather forceful ...
"A ferret's sneeze, which is very loud and sounds like a combination of a cough and a sneeze, may be alarming to the humans in the house."
And, interestingly [?] "At the Centers for Disease Control and Prevention, ferret sneezes are frozen in tiny vials and locked up in a high-security chamber called an enhanced biosafety Level 3 laboratory. ..." (NYT 2006.03.28)
ugis: OK, I'm not sure I understand your point completely. If the flow were vertically (smoothly) laminar then we don't expect a horizontal vectorial force. That includes what happens during a sneeze. The vertical vector would have to far outweigh the horizontal one to prevent horizontal travel before hitting the cage floor (or ceiling). To keep the flow laminar you have to stay under a certain velocity. With high velocity, laminarity can be lost leading to turbulent flow. If the issue is that sneezes shouldn't be classified as airborne but as "contact," fine. I take your point, although I don't think it affects the paper.
Enhanced BSL3 labs are essentially the same as BSL4 labs and refer not to the cabinets but the kind of biosafety procedures used in them and their isolation techniques. Frozen sneezes are really frozen specimens. The sneezing has nothing to do with it. It is the agent that is in the sneeze that is at issue.
revere, Most PNAS are only psuedo-reviewed. They are communicated by an Academy Member who sends the paper out to a couple of buddies, who write up reviews, which are then submitted to PNAS, signed sealed and delivered.
PNAS then makes sure it doesn't exceed the five page limit and the Academy Member is in good standing and hasn't exceeded the "Communicated by" limit.
This paper was "Communicated by".
Henry: I believe that used to be the case (and was when you published there) buthasn't been for some time. PNAS no longer requires an Academy member to communicate a paper and I believe the peer review process is now like that of other journals. I could be mistaken but that is my impression (that tthe process has changed substantially).
OK -- in reverse order:
1. My ref to NYT / CDC-bottled sneezes -- forgetaboutit -- I thought it was cute; but, it's just a distractor
2. What's "airborne"? -- a core question, especially in discussions between aerosol scientists and infection docs, with the (respective) terminology a source of much frustration in their communications. I was just trying to ascertain what YOU meant by airborne. We can leave that, for now.
3. Vertical air flow -- velocity -- near as I can gather from one pub'n, this system is designed to operate at a (spatial) average velocity around 4 feet per minute. But, my use of the word "velocity" was ill-advised; I should have used something like "translocational" or "directivity" i.e. net effect consistently in one direction.
4. Vertical air flow -- directivity -- So, if you've got 2 ferrets separated side-ways by a perforated partition, and respirable particles might otherwise diffuse / tumble through, the vertical air flow will prevent / minimize this. [don't make me do geometry, eh? :-) ]
Sneezes aimed in the right direction are propelled through the partition.
revere, PNAS uses both systems (a real review and a buddy system). If the buddy system is used, the article has "Communicated by" in the heading. The PNAS being discussed is "Communicated by"
see instructiosn below:
In addition, an Academy member may communicate for others manuscripts that are within the members area of expertise. Before submission to PNAS, the member obtains reviews of the paper from at least two qualified referees, each from a different institution and not from the authors institutions. The names and contact information, including e-mails, of referees who reviewed the paper, along with the reviews and the authors response, must be included. Reviews must be submitted on the PNAS review form. The communicating member should include a brief statement endorsing publication in PNAS. In the journal, these papers are distinguished as Communicated by the responsible editor.
Just to clarify the review process by PNAS. If it is a real review, the article says "Edited by". If it is the buddy system, but the Academy member is not an author, then it is "Communicated by". If it is the buddy system and the Academy member is an auther, then it is "Contributed by".
The PNAS paper being discussed was "Communicated by Peter Palese".
Thanks, revere, for your additional input. We at the fluwikie forum had a lot of fun with this paper, from what the heck did they actually discover, to why on earth did the CDC think this deserved all that publicity.
These were some of my initial comments the day the paper came out. 2 days later, they are still pertinent, IMHO, so I am including them here:
They created reassortants with
1) avian internal genes with human HA and NA
2) human internal genes with avian HA and NA, and
3) human RNP genes (PB1, PB2, PA, NP) with the rest avian genes
What they did NOT do was to use avian HA with/without NA + avian PB1 with the rest human genes, which was the reassortments that happened to create the 1957 and 1968 viruses.
The reason why they didn't do that?
"Certain gene constellations, such as rgVic:486HANAPB1 and rgVic:486HAPB1, which reflect the gene constellations of the viruses responsible for the 1957 and 1968 pandemics, respectively, could not be rescued even after multiple attempts, although the same plasmids were used to successfully rescue other reassortant combinations. These results suggest that certain combinations of genes from the avian and human influenza viruses used in this study may not be compatible for virus viability."
So they tried, and it didn't work. Fair enough.
But since there is a world of difference between laboratory conditions and the complex virus/host/new-host/habitat interactions needed to produce a pandemic strain, those results don't really tell us anything about the viability of naturally formed reassortants with those or other genetic combinations.
The other thing which may have been pointed out here is that they used the 1997 H5N1 strain, which has disappeared after culling, and not any of the more recent H5N1 isolates that are related to currently circulating clades.
In summary, what have we got? The study tells us that:
Artificially generated reassortants from human H3N2 and an H5N1 strain that is NO LONGER in circulation, in combinations that have NOT caused pandemics before, are not easily transmitted between ferrets.
So, as I said the first time I read the paper,
Big Deal.
http://www.fluwikie2.com/pmwiki.php?n=Forum.CentersForDiseaseControl
Although there are certainly more combinations that could be tried, H5N1 is fairly well evolved, and isn't going to "improve" just because a few genes were thrown together. As noted in the paper, some combinations could not be rescued. Chances are these failures are due to the fact that certain combinations simply are not viable.
For crossing species barriers, isolates in swine have more variety. There are H3N2 isoaltes that are triple reassortants, with H3, N2, and PB1 from humans, M, NP, and NS from swine, and PA and PB2 from birds. There are also H1N2 isoaltes that have swapped out the human H3 for swine H1. Other isolates (H1N1 and H1N2) have a human PB1 and seven swine genes. These combinations have emerged in the wild and offer selective advantages.
One the other hand, there have been NO reported human or swine genes in H5N1, pandmeic or otherwise. The failure to form any such combinations says something about what does and doesn't offer advantages.
In the PNAS paper, isolates with all 8 human or all 8 avian genes were more fit, in assays that measure replication or transmission, which is useful information and supports the failure of any reported mammalian genes in H5N1 isolates.
Some combinations work, and some don't. At this time there isn't any data that ANY H5N1 would benefit from a mammalian gene.
H5N1 evolves by incrementally changing its genes. The changes are easily seen in teh sequences at GenBank or Los Alamos and any serious evaluation of the polymorphisms acquired leaves little doubt that the acquisitions are via recombiantion and NOT point mutations.
despite the limited relevance of the experiments, I wonder
why not more such experiments are being done.
(e.g. the same with strains from 2005-2006)
Consider the importance of the pandemic issue !
Are these experiments so expensive ?
Can someone please estimate what that study did cost ?
Do you agree that it's reasonable to spend money
for more such experiments ?
so, do you think the paper was political in its wording
and not just scientifical ?
I mean, suppose the same study about a less important
animal's disease - would they have used another presentation and wording ?
The PNAS paper was on data that was generated several years ago, prior to the isolation of the more current H5N1's. The data was upgraded with more current versions. However, the more current versions transmitted less efficiently than the 1997 isolate, when all 8 avian genes were present. More experimaents can be expected, but thus far swapping in human genes have failed to generate a more fit H5N1, when the more efficient earlier H5N1 was used.
anon: These experiments are clumsy because they are done under enhanced BSL3 conditions, which require scarce lab space and skilled personnel. There is demand for the facilities and the personnel, so the most important cost is "opportunity cost," the fact that you can't do other things instead.
anon: No, I don't think the paper was either political or spin. It is what it is.
Papers like these, when quoted by the media, give the public a false reassurance that it's very hard for the virus to change to a pandemic strain. What's not being discussed is that we have several clades and subtypes in play, each with their own evolutionary characteristics.
In addition, what should bother people is these flare-ups every 2-18 months. Instead of automatically attributing the source of each of these flare-ups with local dead chickens where they can be found, we should be comparing the sequence of say the HA gene, to the human isolates. This might give us a clue as to whether the vast majority of cases are in fact due to people being directly infected by sick H5N1 birds, or some other vector. Some will be for sure; all of them? Possibly not.
Revere wrote: "No, I don't think the paper was either political or spin. It is what it is."
For what my two cents are worth: I agree.
When we hear the media accused of fearmongering and hyping "bird flu," it is worth keeping in mind the mediocre media's response to this paper.
Can you say, "Downplay?"
Perhaps reporters would have been even more celebratory without Dr. Gerberding's insistence that the results were NOT reassuring.
Marissa- Of course, you have a financial interest in keeping people scared of bird flu. What kind of consulting work would there be if companies didn't have to "prepare" for bird flu? Interesting that this kind of research is always soft-pedaled by the Reveres, but every time a handful of people get sick in SE Asia we're supposed to get ready at any moment for the other shoe to drop.
Coulter: For the record, I make little money off bird flu preparedness consulting as you call it; indeed I have probably given the equivalent of 150 hours to our community already, gratis. So, get off it.