Sunday, January 08, 2006

Turkey: wait and see

In a story filed yesterday Helen Branswell of Canadian Press reported that WHO had formally confirmed four cases of human H5N1 flu in Turkey. Two other cases, yet to be formally confirmed by the WHO reference laboratory, are presumptive cases. In addition, almost three dozen more cases are hospitalized in eastern Turkey with symptoms suggestive of the disease.
At least 34 people are hospitalized in eastern Turkey with symptoms compatible with H5N1, said WHO spokesperson Cristiana Salvi, reached in Ankara. Salvi said most are under the age of 15.

Of those, authorities have identified at least 12 as probable cases, said [WHO scientist Michael] Perdue. (Branswell in Canadian Press wire story)
The most interesting details in Branswell's story, however, relate to the early genetic sequencing of the Turkish virus.
Meanwhile, a WHO official revealed a preliminary comparison of viral samples taken from the human cases and from affected poultry suggests the viruses that have infected these children remain closely related to the viruses circulating in chickens.

However, further genetic sequencing is needed before the WHO could rule out the possibility that the cases might have been triggered by a mutation that allows the virus to transmit more easily from birds to people.

"All the genes haven't been sequenced. But it's good news that at least in the hemagglutinin and neuraminidase, the genes that have been partially sequenced, that they're ... very closely related,'' Michael Perdue, a scientist from the WHO's global influenza program, said from Geneva.

"We feel a little more comfortable because the animal viruses and the human viruses match pretty closely.''

The hemagglutinin and neuraminidase --the H and N in a flu virus's name--are the surface proteins that allow viruses to invade and spread among the cells of a host.
Well, maybe. But our knowledge of what determines host range and virulence is still rudimentary. And as the song goes, in flu genetics, little things mean a lot. "Close" only counts in horseshoes.

As if to underline this, a paper published this month in the Journal of Molecular Biology (J. Mol. Biol. (2006) 355, 1143–1155) by Stevens and co-workers at Scripps, Mt. Sinai in NY and the Armed Forces Institute of Pathology (Taubenberger) used a new glycan array technology to show that the avian-human species barrier in the 1918 virus was breached by one or perhaps two amino acid changes. A little explanation is in order to understand what they found. If you want still more explanation, you'll find them at The Flu Wiki in our Influenza Primer II.

The first step in cell infection by the influenza virus the binding to a respiratory tract cell's surface by a special protein "key" on the viruses surface. The viral protein in this case is the HA, or hemagglutinin protein, which is the H-part of the usual H-N designations of influenza viruses. There are 16 broad types of HA proteins (H1 - H16), of which H1, H2 and H3 are found on human adapted flu viruses (NB: when discussing the hemagglutinin protein generically it is usually referred to by HA, but when the subtypes are at issue, the "A" part is dropped and a number substitute; thus H5 is a subtype of the HA protein of the virus).

The species specificity of the virus (its host range) is thought to be strongly related to whether the HA protein binds to one of two different receptors, designated the α-2, 3 or the α-2, 6 linkages (I am leaving out a lot of details, but this will give you the idea). α-2, 3 linkages are found in the intestines of birds, the α-2, 6 linkages in the lungs of humans. Pigs have both and for some time the theory was that pigs could hold both human and bird viruses and provide the "mixing vessel" for reassortment of bird and human viruses. But we now know that humans also have α-2, 3 linkages on the cells of the upper respiratory tract, which might explain why heavy exposures to bird viruses are capable of infecting humans, too, as we have learned in recent years where human cases of influenza with H5, H7 and H9 sub-types have been described. The current Asian/Eurasian outbreak in people is an H5 subtype.

What the Stevens paper did was examine the predilection for either α-2, 3 or α-2, 6 of various H1 or H3 viral HAs, i.e., human adapted viruses and their avian counterparts. In particular, the two of the H1s were proteins from two 1918 soldiers that died on the same day during the second wave of the disease, September 1918, each within 6 days of hospitalization. One died in South Carolina and one died in New York. They were clinically similar but it turns out the HAs of the viruses were not the same, and both were very close to the bird consensus sequences.

Indeed, the 1918NY virus differed in only one amino acid from the bird sequence, while the 1918SC virus had an additional mutation. The two 1918 HAs differed also in receptor predilection. The 1918NY bound to both α-2, 3 and α-2, 6 while the 1918SC virus bound mostly to α-2, 6. In either case, single mutations within the receptor binding site caused substantial and significant alteration in the host specificity of the virus, from bird to human.

Clearly this is not the entire story, as there are other variations in the combination of specificity and virulence not explained by this relatively simple scheme. What it does suggest, however, is that we can't be too confident that viruses that are "very close" to the poultry virus are in fact also very clsoe to it epidemiologically or clinically.

Said another way, I am not particularly reassured by the sequence data. The proof of the pudding will be what happens on the ground in Turkey and elsewhere. At the moment that is not completely clear, although the histories of the confirmed cases suggest heavy exposure to infected poultry.

Before "breathing easier" I prefer to wait and see what we find out over the next week or two.