Commentary
Evolution of H274Y Tamiflu Resistance Baffles Experts
Recombinomics Commentary 23:55
July 13, 2008
The preliminary analysis suggests that the resistant mutants do not share a single origin and further genotypic analysis is ongoing.
The above comment from the WHO update entitled "Influenza A (H1N1) virus resistance to oseltamivir" is confirmed by a multitude of phylogenetic trees of recent H1N1 isolates from a number of countries, which can be analyzed in conjunction with public sequences. These sequences contain H274Y, which confers Tamiflu resistance. The sudden appearance of Tamiflu resistance in seasonal flu linked to independent introductions of H274Y into populations that rarely use Tamiflu has startled and baffled influenza “experts” who expect such antigenic drift to be linked to selection of random mutations. However, the explosion of Tamiflu resistance in seasonal flu is just one of many surprises associated with this change, which is chronicled below.
H274Y gained considerable attention when it appeared in Tamiflu treated H5N1 patients in Vietnam in 2005. H5N1 was linked to an explosion of human cases in Vietnam and Thailand in 2004 and the H5N1 was classified as clade 1, in contrast to clade 2 H5N1 in birds which had emerged in China and multiple countries to the east of China. Clade 1 was a concern because it was quite different from the earlier H5N1 outbreaks in humans and stockpiled vaccines were a poor match. Moreover, all clade 1 isolates had two amantadine resistance markers in the M2 protein, so the only anti-virals of choice with the neuraminidase inhibitors. However, Relenza was not available in quantity, so the drug of choice was Tamiflu (osletamivir).
However, earlier data on use of Tamiflu against the nine neuraminidase serotypes indicate N1 was one of the serotypes with the highest level of resistance. Moreover, inhibition testing of H5N1 from Vietnam required an even higher level of Tamiflu to achieve a 50% marker, indicating treatment dose levels were near the limits of efficacy. Use was compromised further because Tamiflu was most effective when administered with 48 hours of disease onset, and in most cases treatment would begin many days after disease onset because H5N1 infections initial present with many of the same symptoms as seasonal flu, which is not traditionally treated with anti-virals in southeast Asia.
However, Tamiflu was the only real anti-viral choice, so it was used to treat and prevent H5N1. The prophylactic dose was half of the treatment dose, raising additional concerns of resistance due to sub-optimal treatment, but its use was implemented in a plan to blanket H5N1 regions with Tamilfu, to blunt and initial human transmission.
In 2005, Tamiflu resistance was identified in H5N1 patients being treated with Tamiflu, as well as one contact who was receiving a prophylactic dose because her brother was hospitalized and H5N1 positive. H274Y was identified in patients being treated as well as the patient being treated initially with a prophylactic dose. Moreover, the prophylactic patient had H5N1 with a second Tamiflu resistance change N294S. Moreover, H274Y was also detected in a member of the Karo cluster in Indonesia who had prematurely stopped taking Tamiflu. In Indonesia, the H5N1 was clade 2.1.
However, assurances were given that the resistance markers, especially H274Y, was associated with a fitness penalty, so although the resistance would cause problems in infected patients, it would only be viable in patients that were being treated with Tamiflu, because H274Y would reduce the ability of H5N1 to compete with wild type H5N1. However, the fitness penalty was previously reported in seasonal flu, and the applicability to H5N1 was unclear.
There were several pieces of data that raised concerns about assurances based on a fitness penalty. Both H274Y and N294S had been previously been reported in poultry suggesting that H5N1 with either mutation could compete with wild type H5N1. Additional concerns were linked to a cluster of H5N1 cases in Egypt, which had N294S in samples collected prior to Tamiflu treatment. Moreover, H274Y was detected in wild bird isolates in Astrakhan, raising additional doubts about a fitness penalty. The H5N1 in the above instances was clade 2.2 extending the number of different H5N1 clades and sub-clades with H274Y or N294S.
Concerns that the Tamiflu resistance could be acquired by seasonal flu began shortly thereafter in the 2006/2007 season, when five N1 sequences with H274Y became public. All five H1N1 infected patients were from the United States, but these five sequences represented less than 1% of N1 sequences that season, so the significance was unclear, although H274Y was not reported for any prior human H1N1 public sequence.
Concerns grew markedly the followings season, when Norway reported that more than half of the H1N1 infections contained H274Y. These concerns increased when H1N1 sequences from the United States were released. Approximately 30% of the Brisbane/59 (clade 2B) had H274Y, and there were at least independent introductions. Recently phylogenetic trees identify an number of additional introductions in isolates from patients in Japan which had not been treated with Tamiflu prior to sample collection. The data from other countries indicated the H274Y was concentrated in clade 2B, and the frequencies were increasing in multiple countries.
This global spread of H274Y through introductions onto multiple gentic backgrounds indicated the H274Y was not spreading via selection of random mutations, but instead was being acquired by homologous recombination as was previously described for G743A in H5N1.
The data support the concurrent acquisition of the same polymorphism onto multiple genetic backgrounds and suggest Tamiflu will have limited utility in the treatment of prevention a pandemic linked to efficient transmission of H5N1.
Media Links
Recombinomics Presentations
Recombinomics Publications
Recombinomics Paper at Nature Precedings