Paradigm Shift Intervention Monitoring
twitter
Live feed of underlying pandemic map data here
Commentary
Potential Pandemic
H1N1 Vaccine Mismatches Raise Concerns
Recombinomics
Commentary 11:42
March 15, 2010
The above comments are from the CDC's MMWR report on initial analysis of seasonal H1N1 in the 2007-2008 season based on a poor reference anti-sera that failed to identify low reactors. The above isolates were actually Brisbane/59/2007-like isolates which were designated low reactors when appropriate anti-sera was used. The target was changed to Brisbane/59 the following season. However, Tamiflu resistance gained a foothold in Brisbane/59 that season, and went to 100% in the following season.
Parallels with the 2009-2010 pandemic H1N1 season are striking and of significant concern.
Recently released HA sequences (at GISAID) by CDC have G158E, which has been linked to low reactor status. Two German isolates, A/Bayern/62/2009 and A/Bayern/69/2009, each have only one recent non-synonymous polymorphism, which is G158E. Mill Hill designated both isolates as “low reactors”. One of the two was also tested by the CDC, and it was again designated a “low reactor”. Analysis of isolates that escaped from neutralizing monoclonal antibodies also had G158E, as well as changes at the adjacent upstream position (K157E and K157Q). Similarly, the first two US low reactors identified by the CDC each had 4 changes, but in both cases one of the four was N159D, which is at the adjacent downstream position. All three positions (157-159) map to the same antigenic site, strongly supporting the involvement of G158E in immunological escape, as indicated by a “low reactor” status.
However, although the identification of G158E has been made by multiple independent approaches involving three labs, including the CDC, the CDC antigenic characterization of these more recent isolates did not have the “low reactor” designation. Recently WHO has acknowledged inter-lab variations and has produced an international standard using pooled human anti-sera. Although it is not clear if the absence of low reactor status for recent isolates is due to the use of the international standard, or some change or variation in the CDC assays, the failure to detect low reactors has been detailed previously, as noted above.
In the 2007/2008 season a new H1N1 target was selected. Previously, the target was New Caledonia/20/1999, which was clade 1 and the CDC correctly predicted that New Caledonia would be replaced with clade 2, and Solomon Islands/3/2006 was selected as the target. However, by fall Solomon Islands/3, which was clade 2A, had been replaced by two other sub-clades. Brisbane/59/2007 (clade 2B) and Hong Kong/2562/2006 (clade 2C). The CDC had developed a ferret anti-sera against Solomon Islands/3 which could distinguish clade 1 from clade 2, but could not distinguish between the three clade 2 sub-clades. Consequently, as noted above, all isolates were called Solomon Island-like, even though Solomon Islands/3 was not circulating. However, other labs had antibodies that could better distinguish the three sub-clades and soon these other sub-clades were designated low reactors. Using an 8 fold reduction in titer as the cut-off, both Mill Hill and Australia were finding that approximate 46% and 45%, respectively, of H1N1 isolates were low reactors while the CDC was only finding 4% (see slide 7). However, phylogenetic analysis showed that both HA (see slide 8) and NA (see slide 9) had significant non-synonymous changes when either clade 2B or clade 2C were compared to clade 2A or each other, while clade 2B differences between H1N1 isolates from the United States or Europe were minimal (see isolates with LR label in phylogenetic trees), so data generated by the CDC should have been similar. Moreover, the minimal differences between clade 2B isolates suggested the number of isolates showing a four fold reduction in titer would be similar to the number giving an 8 fold reduction (the difference was due to a 2 fold variation in the assay), and the number of low reactors in clade 2B was close to 100%. In addition the appearance of Tamiflu resistance was highest in clade 2B, so the vaccine mismatch may have contributed to the fixing of H274Y.
The phylogenetic data was confirmed when Australia then made two anti-sera against Brisbane/59. One immunization used Brisbane/59 grown in eggs, while the other used virus from mammalian cells. The egg based anti-sera extensively cross reacted with the three clade 2 prototypes, but the anti-sera against Brisbane/59 grown in mammalian cells gave very clear cut results. The titer against clade 2B was 320, but fell to 40 (8 fold) when tested against clade 2C, and gave now reaction (titer less than 40) when tested against clade 2A or clade 1 (see slide 5). Thus, the antigenic characterization data supported the phylogenetic data which predicted major differences between the three clade 2 sub-clades, and anti-sera against Solomon Islands/3 would offer minimal protection against Brisbane/59-like viruses of Hong Kong/2652-like virus. Thus, although the target was a mismatch because the wrong clade 2 target was selected, initial characterizations published in MMWR claimed a match by stating that the vaccine was directed against Solomon Islands/3 and all 2007 isolates in the 2007/2008 season were Solomon Islands/3-like. Thus, the CDC data was really generated from an answer (match) in search of an assay.
There is considerable concern that the answer in search of an assay approach is being applied to pandemic H1N1. California/7/2009 also appears to be a mismatch or is close to a mismatch. It has 5 amino acid differences with the consensus H1N1 sequence, so although the consensus is not labeled a mismatch, single changes can push the H1N1 into the low reactor category. This happened when Mill Hill declared a Ukraine isolate, A/Lviv/N6/2009 a low reactor. It had only one additional HA change, D225G. Similarly, two German isolates which just had G158E were also designated low reactors by Mill Hill, and one, A/Bayern/59/2009 was also tested by the CDC and also designated a low rector. Concerns about an answer in search of an assay were raised when the CDC tested an isolate from the Lviv/N6 patient. The CDC isolate had both G158E and D225G yet the CDC said it was not a low reactor raising serious concerns about sensitivity of the assay. Similarly, US isolates with G158E were also not labeled low reactors by the CDC, suggesting the assay had changed. At about the same time the WHO announced the creation of an “International Standard” created by pooled anti-sera from human patients. The use of pooled anti-sera also raises questions about an answer (no 4 fold drop in titer) in search of an assay.
Thus, the parallels between H1N1 seasonal flu antigenic characterization tests and associated anti-sera in 2007/2008, and inter-lab variations in pandemic H1N1 in the 2009/2001 season, raise serious concerns that the failure of the CDC to designate samples with G158E or D225G as low reactors is more closely related to the assays being used rather than the actual ability of the 2009/2010 vaccine to effectively recognize such variants. The concerns were increased by vaccine failures in the US and Europe.
These failures are clearly hazardous to the world health.
Media Links
Recombinomics
Presentations
Recombinomics
Publications
Recombinomics
Paper
at Nature Precedings