Migration of Uva Lake H5N1 to Europe and the Middle East
Recombinomics Commentary
December 18, 2007
"It is live birds, not dead birds, that shed the virus and pose a threat. Because our surveillance efforts have struggled to find infected live birds in non-endemic regions, it is essential that we shift sufficient resources and efforts to evaluate H5N1 transmission in wild birds in known endemic areas such as southeastern Asia, China, Indonesia and Africa."
I am reminded of a class given me in 1964 by my Tulane epidemiology professor Ken Newell on the praticalities of investigating an outbreak and deciding where to go 1st. He strongly advised going 1st to look at the outlying cases because the spread factors are few there and probably are among the most important. In the centre of an epidemic one is presented with a score or more of possible explanations and no easy way of sorting them out. His advice has helped me many times. With the insights from the outer investigations one is better prepared when diving into the middle. However Walter Boyce's comments are very opportune and well worth considering in these confusing times. - Mod MHJ]
The above comments, from the most recent "Avian Influenza: Poultry vs Migratory Birds" Promed report is curious, because the suggestion for looking at outlying cases was done in the summer of 2005. The results were unequivocal, yet it was after the results were in, that Promed began a series of articles related to “dead birds don’t fly” and “wild birds as victims” and each time hard data was presented to further support the obvious involvement of wild birds in the transport and transmission of H5N1, a series of hand waving arguments using negative data would follow, and this pattern has continued for well over two years, including comments on the recent H5N1 outbreak in England.
The strongest hand waving began shortly after H5N1 was confirmed at Qinghai Lake in central China in May, 2005. That outbreak began with 178 dead bar-headed geese in early May, By the time China filed its OIE report a couple of weeks later, the number of dead birds reported had grown to 519. Although most of the dead birds were bar-headed geese, other geese, gulls, and cormorants were included. Qinghai Lake is the largest lake in China and many species arrive there in the spring, so the finding of H5N1 in multiple species raised concerns that long range migratory birds could spread H5N1 far and wide. Bar-headed geese can fly 1000 miles in 24 hours and Qinghai Lake is located at the intersection of multiple major flyways.
However, the “dead birds don’t fly” argument, held that the lethality of the H5N1 would cause it to die off at Qinghai Lake and the virus would not spread. The H5N1 at Qinghai Lake was novel (Clade 2.2) and in pathogenicity studies, it killed experimental chickens within 24 hours. Similarly, experimental mice were killed within 3 days, which was likely linked to the fact that all isolates had PB2 E627K, which was a mammalian polymorphism that allowed the virus to replicate more efficiently at lower temperatures and was linked to increased virulence in mammals. Similarly, H5N1 infections usually did not cause death in waterfowl, and the positives were dead waterfowl. Moreover, at that time the “Asian” strain of H5N1 had not been reported in any country west of China, including Mongolia and southern Siberia where many flyways intersect and long range migratory birds summer. However, although the number of dead birds ranged from 5,000 to as high as 10,000 and most were bar headed geese, many birds and species clearly survived the outbreak.
In July the “dead birds don’t fly” argument was destroyed when Russia reported H5N1 on farms surrounding Chany Lake in southern Siberia followed by H5N1 outbreaks across the border in Kazakhstan. Sequencing of the H5N1 showed that it was the Qinghai strain and the distance from Qinghai Lake voided the “dead birds don’t fly” explanation. The pattern of infection surrounding the lake clearly pointed toward the wild birds as the source of the infection but the corollary of “wild birds as victims” was invoked, arguing that the wild birds had been infected by the poultry.
The basis of this argument primarily rested on prior observations that both H5 and H7 infections can begin as low path infections on poultry farms, and then evolve into high path by acquiring a polybasic cleavage site via non-homologous recombination, which might be facilitated by the dense concentration of susceptible hosts at poultry farms. However, in the case of “Asian” H5N1, this was not the case because the polybasic cleavage site was identified in Guangdong province in 1996, and the cleavage site was fixed in wild birds, domestic poultry, and infected mammals, including humans. The cleavage site of the Qinghai strain was a polybasic cleavage site with minor variations on the theme. Similar cleavage sites were found in various H5N1 “Asian” sub-clades (1, 2.1, 2.2, 2.3).
“Dead birds don’t fly” and “wild birds as victims” were simultaneously destroyed by the isolation of H5N1 from a healthy crested grebe at Chany Lake. The sequence was clearly the Qinghai strain and it was virtually indistinguishable from H5N1 in the domestic poultry or dead wild birds. Thus, the Qinghai strain required no modifications by domestic poultry and could be isolated from healthy wild birds.
However, the beginning of the negative data campaign began at Erhel Lake in Mongolia, also in August of 2005. Wild birds had begun dying there also. Since Erhel lake is remote and far from any poultry farms, identification of Qinghai H5N1 would settle any remaining issues on the transport and transmission of H5N1 by migratory birds. However, this investigation was aided by a wildlife conservation groups, and comments to the media and press releases set the tone for the upcoming campaign.
Initially, the dead wild birds were acknowledged, but the role of H5N1 was discounted because there were not enough dead birds. When H5 was found in a bar headed goose and whooper swans, H5N1 was again discounted because not enough birds were dead. When high path H5N1 was identified in these birds, it was discounted because H5N1 was not detected in the live birds at the lake.
The failure to find H5N1 in live birds at a remote lake, when H5N1 was found in dead wild birds, clearly demonstrated that the live bird assay was fatally flawed. Live birds carried the Qinghai strain to the remote lake, and domestic poultry was not required to modify the genetic information to convert low path to high path. Like Russia and Kazakhstan, Mongolia had also never reported “Asian” H5N1, so all three countries represented outliers, with new introductions of H5N1 into wild bird populations. Thus, the suggestion in the Promed commentary was followed, and there was little doubt on the science side that H5N1 had moved into the long range migratory birds population at the intersection of major flyways, which would lead to the spread of H5N1 into Europe, the Middle East, and Africa.
This prediction was strengthened by further data from Russia in the summer of 2005 (the Mission report). Testing of hunter killed healthy wild birds identified H5N1 in over 25 species. Thus, as expected, infection was not limited to a small number of species and transport did not require movement by a single bird or a single species. The large number of birds and species would facilitate the spread and the predictions came to fruition in the fall of 2005.
H5N1 was first reported in the Volga Delta, signaling movement south as wild birds began to migrate out of Siberia and Mongolia. Like the earlier outbreaks, the “Asian” strain of H5N1 had never been reported west of China. Sequence analysis characterized the H5N1 from mute swans as the Qinghai strain. That was followed by more wild bird positives in Romania in the Danube Delta and in Ukraine on the Crimea Peninsula. In each instance, the H5N1 was the Qinghai strain.
However, one of the most important discoveries was H5N1 in a healthy teal in the Nile Delta in December, 2005. The H5 was discovered in a PCR test, but the researchers at NAMRU-3 couldn’t get the virus to grow. Consequently, they had to run multiple extractions to get enough RNA for a sequence. The HA and NA sequences were clearly the Qinghai strain and were subsequently shown to be virtually identical to H5N1 isolates in Austria in 2006. Thus, in December, 2005 H5N1 had already migrated into the Nile Delta, but at that time every EU country in Europe was denying the presence of H5N1, as was every country in Africa. Even Egypt did not have reported H5N1 until 2006, because of difficulties in extracting and sequencing the H5N1 from the healthy teal.
At that time, the same conservation groups that couldn’t detect H5N1 in live wild birds at Erhel Lake cited the detection failures in Europe as evidence that dead birds didn’t fly, even though H5N1 had been reported at the Volga Delta, Danube Delta, and Crimea Peninsula. However, at the beginning of 2006, H5N1 was isolated from fatal human infections in Turkey, which was then followed by admissions of widespread H5N1 in Turkey in December and possibly November in both domestic and wild bird. Neighboring countries in the Middle East continued to deny H5N1 infections, but began to cull birds neighboring Turkey. In February an H5N1 cluster was confirmed in victims neighboring Iraq along with wild and domestic birds in many countries in Europe, the Middle East, and Africa. Human cases were reported in Azerbaijan, Egypt, and Djibouti. In all instances the H5N1 was the Qinghai stain.
In Europe, most countries reporting H5N1 were from wild bird infections, although the detection was exclusively in dead or dying birds, Only Russia and Egypt reported H5N1 in live healthy wild birds. Countries in Europe maintained that there was no H5N1 in EU countries in 2005, and a harsh winter in early 2006 led to a westward migration of birds from eastern Europe. However, this explanation was not supported by the sequence data which showed the H5N1 in Austria in early 2006 matched H5N1 in the Nile Delta in 2005. Similarly much of the H5N1 in EU countries in early 2006 did not match the regional markers in eastern Europe, Although all were the Qinghai strain, the sub-clade sequences supported independent introductions by wild birds throughout Europe, the Middle East, and Africa. In each of these areas, H5N1 was detected in wild birds, but in dead wild birds.
The detection failures in live wild birds led to further “dead birds don’t fly” and “wild birds as victims” campaigns. The conservation groups tested hundreds of thousands live birds with assays that had been fatally flawed in Mongolia to generate negative data for Europe, the Middle East, Africa, and North America, even in areas where H5N1 is found in dead or dying wild birds.
Surveillance shortfalls also became clear in the 2006 / 2007 season. In the summer of 2006 there was a massive die off of wild birds at Uva Lake, the largest lake in Mongolia. The die-off was on a par with Qinghai Lake, a year earlier. The H5N1 at Uva Lake was the Qinghai strain, but it was an easily distinguishable sub-clade. However, in the fall of 2006 and beginning of 2007 no country in Europe reported detecting the Uva Lake strain. The winter in 2006 /2007 was milder in Europe and reported infections were markedly lowered. At the Options VI influenza conference in Toronto, in June, 2007, H5N1 in wild birds in Europe was said to be virtually eliminated. The large number of negative live birds was cited as evidence that wild birds were not a major threat, and H5N1 spread was due to trade and illegal smuggling.
However, within hours of the talks by researchers in Italy and England, H5N1 was reported at a poultry farm in the Czech Republic. That report was quickly followed by H5N1 detection in wild birds in the Czech Republic and neighboring Germany. The outbreaks in Germany were widespread and included Bavaria, Saxony, Saxony Anhalt, and Thuringia. Over 300 wild birds were H5N1 confirmed, approaching levels detected in early 2006, when Germany reported the highest number of H5N1 confirmed wild birds in Europe. Once again H5N1 was widespread in Europe, but the detection was in the middle of the summer, when long range migration was minimal, indicating H5N1 had become endemic in wild birds in Germany, even though there were virtually no reports of H5N1 between the summer of 2006 and summer of 2007. Moreover, the H5N1 detected in Germany was the “missing” Uva Lake strain. Prior to the summer of 2007, the only country reporting this strain in 2007 was Kuwait.
However, in September 2007 the sequence of H5N1 from a chicken outbreak in Krasnodar was published, and it was clearly Uva Lake, as were sequences from three regions in Germany. A wild bird (whooper swan) sequence from Krasnodar was also published and the HA sequence was an exact match of the chicken sequence and the two isolates were 99.95% identical. The Krasnodar isolates were also Uva Lake but like the sequences from three regions in Germany, represented independent introductions that had identities above 99%, but well shy of the greater than 99.9% identity seen when the isolates were from a common source.
Reports also indicate that the outbreaks over the summer in the Czech Republic and France were the Uva Lake strain as was the more recent outbreak in England. The outbreak in England involved free range turkeys, and the Royal Bird Watchers in England were again citing negative data generated by the dismal surveillance program in England, which not only failed to find H5N1 in live birds, but has only found H5N1 in one dead swan on the shores of Scotland, although H5N1 has been repeatedly detected in dead wild birds in Germany, Denmark, Sweden, and France. Also cited were detection failures in western Europe between the summer of 2007 and the outbreak in England as evidence that the Uva Lake strain was due to an unreported domestic poultry reservoir somewhere in Europe. The argument by the bird watchers ignored the outbreak in Krasnodar in September and the more recent outbreaks in the Danube Delta in Romania and multiple locations in northern Poland, which included H5N1 in wild birds. Moreover, the most recent outbreak reported in Europe is in northeastern Germany, adjacent to Poland, and as indicated yesterday, the German outbreak is due to the Uva Lake strain.
Thus, the Uva Lake strain in the fall of 2006, like the Qinghai Lake strain in the fall of 2005, went undetected in western Europe, again raising serious surveillance questions relative to dead wild birds. Moreover, the detection failures in live wild birds highlights a fatally flawed program that uses methods and assays which lack sensitivity to detect H5N1 in live wild birds.
These flaws extend worldwide, especially to countries in Europe, the Middle East, and Africa which fail to find H5N1 in wild birds prior to outbreaks in domestic poultry. These countries include Saudi Arabia, where over 4 million birds have been culled just before the start of the Hajj, and Pakistan where a large human cluster will be confirmed today, with an obvious human-to-human transmission pattern.
The surveillance failures remain hazardous to the world's health.
Media Links
Recombinomics Presentations
Recombinomics Publications
Recombinomics Paper at Nature Precedings