Commentary
The
Emergence of H5N1 Tamiflu Resistance
Recombinomics Commentary
17:11
April 1, 2009
"The honest truth is no one knows," said Osterholm, director of CIDRAP.
However, he told the group that the current level of antiviral resistance won't necessarily persist, as sensitive flu strains might replace resistant ones. Osterholm also pointed out that antiviral resistance hasn't been detected in influenza A/H3N2, a previous pandemic strain that is now a common seasonal strain.
The above comments fail to consider the wealth of sequence data that has recently emerged, in part because of concerns regarding antiviral resistance in H3N2 and H1N1 seasonal flu. Adamantane resistance in H3N2 has now become fixed (at or near 100%), as has resistance in clade 2C of H1N1. Similarly, osletamivir resistance has become fixed in clade 2B of H1N1. Sequence data supports the fixing via genetic hitch-hiking that is not linked to antiviral usage. This hitch-hiking via homologous recombination has been most dramatic for H274Y (the change conferring Tamiflu resistance) in H1N1. Moreover, since H274Y has also been reported in H5N1, the jump of H274Y from H1N1 to H5N1 is a serious concern, that is now more likely since H274Y has become fixed in clade 2B, which has become the dominant sub-clade of H1N1 worldwide.
The independence of H274Y emergence from Tamiflu usage has been demonstrated via a number of lines of evidence. Resistance to Tamiflu via selection leads to changes at multiple positions in N1 in H1N1 as well as N2 in H3N2. However, there have been no recent reports of resistance at any position other than H274Y in H1N1. Similarly, last season countries that used oseltamivir frequently, such as Japan, had low levels of H1N1 resistance (3%), while counries that used oseltamivir at low levels, such as Noway, had high levels of resistance (67%). Moreover the patients that were infected with resistant H1N1 were not taking Tamiflu.
In addition, most of the resistance was found on a single sub-clade within clade 2B, and expansion of this sub-clade led to fixing the previous season. However, prior to the fixing, H274Y jumped from sub-clade to sub-clade (2C to 1 to 2B) and from branch to branch within each sub-clade. Thus, there were multiple introductions of H274Y into hosts that were not taking Tamiflu. Such introductions are most easily explained by homologous recombination.
Suppport for homologous recombination was also provided by the acquisition of key changes in NA and HA that led to the fixing of H274Y in H1N1. These included three sequential polymorphisms in NA that had been previously found in clade 2C. Similarly, the key change in HA, A193T, also was in clade 2C prior to jumping to clade 2B. The jumps of multiple polymorphisms from 2C to 2B on two different gene segments and the inclusion of three consecutive changes provides compelling evidence for homologous recombination driving these key changes, which are not easily explained by a copy error / selection mechanism, especially when acquisitions are synonymous (do not change the protein sequence), or are for resistance, but found in patients not taking Tamiflu.
Acquisition via recombination has serious implications for the emergence of H274Y in H5N1. Although acquisition by reassortment and recombination require dual infections involving H1N1 and H5N1, the acquisition of a human flu gene by H5N1 has never been reported. In contrast, the jumping of H274Y from one sub-clade to another has been demonstrated multiple times, as described above, and the identical change has been reported in H5N1 in wild birds, demonstrating that such an acquisition does not create a significant fitness penalty in H5N1, as has also been demonstrated in multiple sub-clades in H1N1 seasonal flu.
Moreover, the fixing of H274Y in H1N1 increases the likelihood of productive co-infections. In addition, the likelihood that the clade 2B with H274Y will be displaced by another sub-clade is low, in part because of an outdated vaccine target selection procedure. That procedure relies on unreliable ferret anti-sera results, which leads to selection of vaccine targets which produces vaccines that are outdated and have limited utility.
These deficiencies were readily seen in recent H1N1 vaccine targets. Last season the vaccine target was changed from clade 1 (New Caledonia/20/1999) to clade 2A (Solomon Island/3/2006). However, last season clade 2A was no longer in circulation. Instead, it had been replaced by clade 2B (Brisbane/59/2007) and clade 2C (Hong Kong/2652/2006).
However, the H1N1 last season was characterized with ferret anti-sera generated against Brisbane/59 grown in chicken eggs, which produce extensive cross reactivity with all three clade 2 sub-clades, and these sub-clades were designated as being "antigenically indistinguishable". However, the three sub-clades were easily distinguished through phylogentic analysis, due in part to a large number of non-synonymous changes, including those near position 190. Subsequent anti-sera directed against Brisbane/59 grown in mammalian cells readily distinguished the three clade 2 sub-clades, demonstrating that there were indeed antigentically distinguishable, leading to lower activity for the vaccine that was directed against clade 2A, which was not in circulation. Thus, the vaccine mismatch of last season would have enhanced the emergence of the two H1N1 sub-clades in circulation, which were not targeted.
This season, the H1N1 target was changes to clade 2B, but the target was Brisbane/59, which was isolated in the summer of 2007. By late 2007 it was already clear that the Tamiflu resistance was linked to a Brisbane/59 sub-clade that contained a number of key changes that were not in the vaccine target, including NA changes such as H274Y and the three sequential polymorphism that had been previous been in clade 2C.
Moreover, in late 2007 a sub-clade within the sub-clade began to emerge that had the above changes as well as A193T in HA. This acquisition was important because it also had been in clade 2C previously, as well as H1N2 that had been in circulation in 2003. Thus, A193T was associated with significant prior seasonal flu sub-clades, and this sub-clade had already spread to multiple locations in North America and Europe prior to the vaccine selection in February 2008.
Thus, although the dominant strain was clade 2B this season, the clade 2B had a number of acquisitions that were not present in the 2007 target. The number fo difference increase over the summer, when Tamiflu resistance rose to 100% in South Africa and approached 100% in other countries in the southern hemisphere, such as Australia and the Philippines. The dominant H1N1 in these countries again had A193T, which was flanked by additional changes at positions 187, 189, and 196. However, none of these changes were incorporated into the H1N1 vaccine target, which again was Brisbane/59 for the 2009 season in the southern hemisphere.
Similarly, these changes were also present in isolates from the beginning of this season, but the number of changes increased. There were two polymorphisms at 187 (N187S and N187D), two at position 196 (H196R and H196N), and four at position 189 (G189S, G189A, G189V, G189N) along with A193T, which was in all of the isolates. Although the presence of A193T was noted in the report on vaccine selection for 2009/2010, the H1N1 vaccine target remained unchanged, even though multiple countries with G189A noted vaccine resistance.
Consequently, H274Y has now paired up HA changes at A193T and a number of flanking changes, which will likely lead to the emergence of one or more of these combinations. This emergence is facilitated by a lack of these changes in any of the current and future clade 2B vaccine targets, which are all the Brisbane isolate from 2007.
Thus, it is likely that H274Y will be circulating in the upcoming season in the southern hemisphere, increasing the likelihood of a co-infection of H5N1 with H1N carrying H274Y and failure to change the target for 2009 or 2010 increases the likelihood that H274Y will continue to in H1N1, so although it is true that no one knows fro certain about H274Y in H1N1 in the near term, or the impact of H274Y in H1N1 on the likelihood of a jump to H5N1, the amount of data associated with such an event has increased significantly in the past 12 months.
However, these data are being actively ignored by those selecting vaccine target and those advising on pandemic preparedness.
Recombinomics
Presentations
Recombinomics
Publications
Recombinomics
Paper
at Nature Precedings