Oseltamivir-Resistant Influenza Viruses A (H1N1) during 2007–2009 Influenza Seasons, Japan

Prevalence of these viruses increased during the 2008–09 season.

I nfl uenza A and B viruses are major pathogens that represent a threat to public health with subsequent economic losses worldwide (1). Vaccination is the primary method for prevention; antiviral drugs are used mainly for prophylaxis and therapy. Currently, 2 classes of drugs, matrix 2 (M2) blockers and neuraminidase inhibitors (NAIs) are available, but M2 blockers such as amantadine and rimantadine are not commonly used because of the rapid generation of resistance and lack of effi cacy against infl uenza B virus (2)(3)(4). The NAIs zanamivir and oseltamivir are widely used because of effects against infl uenza A and B viruses and a low frequency of resistance. NAI virus surveillance studies by several groups have demonstrated that <1% of viruses tested show naturally occurring resistance to oseltamivir as of 2007 (5)(6)(7)(8)(9)(10), indicating limited human-to-human transmission of these viruses.
At the beginning of the 2007-08 infl uenza season, however, detection of a substantially increased number of oseltamivir-resistant infl uenza viruses A (H1N1) (ORVs) was reported, mainly in countries in Europe where the prevalence varies, with the highest levels in Norway (67%) and France (47%) (11)(12)(13)(14). These viruses showed a specifi c NA mutation with a histidine-to-tyrosine substitution at the aa 275 position (N1 numbering, H275Y), conferring highlevel resistance to oseltamivir. Most of these ORVs were isolated from NAI-untreated patients and retained similar ability of human-to-human transmission to oseltamivirsensitive infl uenza viruses A (H1N1) (OSVs) (10,15). In response to public health concerns about ORVs, the World Health Organization (WHO) directed Global Infl uenza Surveillance Network laboratories to intensify NAI surveillance and announced regularly updated summaries of ORV data collected from each laboratory on its website (16). This site reported that the global frequency increased from 16% (October 2007-March 2008) to 44% (April 2008-September 2008) to 95% (October 2008-January 2009), indicating that ORVs have spread rapidly around the world. Japan has the highest annual level of oseltamivir usage per capita in the world, comprising >70% of world consumption (10). Such high use of oseltamivir has raised concerns about emergence of OSVs with increased resistance to this drug. Moreover, in Japan, 2 recent infl uenza seasons were dominated by infl uenza viruses A (H1N1) (Figure 1). If a high prevalence of ORVs is observed, primary selection of oseltamivir treatment for infl uenza patients should be reconsidered. Thus, monitoring ORVs is a serious public health issue.
To estimate the frequency of ORVs and characterize these viruses, we analyzed 1,734 clinical samples isolated from the 2007-08 season and 1,482 isolates from the 2008-09 season by NA sequencing and/or NAI inhibition assay. The total frequencies were 2.6% in the 2007-08 season and 99.7% in the 2008-09 season, indicating that ORVs increased dramatically in Japan.

Sequence Analysis
The phylogenetic tree of NA and HA1 genes was constructed by neighbor-joining methods. The phylogenetic tree was described by representative ORVs and OSVs isolated from several prefectures in Japan. Sequence information for isolates from other countries was obtained from the Global Initiative on Sharing Avian Infl uenza Data and the Los Alamos National Laboratory database. All amino acid positions in the phylogenetic tree were described by N1 numbering.

NA Inhibition Assay
The chemiluminescent NA inhibition assay was per-Emerging Infectious Diseases • www.cdc.gov/eid • Vol. 16 formed by using the NA Star Kit (Applied Biosystems, Tokyo, Japan) with slight modifi cations of the instructions provided by the manufacturer. The fi nal drug concentration ranged from 0.03 nmol/L to 6,500 nmol/L for oseltamivir and from 0.03 nmol/L to 12,500 nmol/L for zanamivir. Chemiluminescent light emission was measured by using an LB940 plate reader (Berthhold Technologies, Bad Wildbad, Germany). Drug concentrations required to inhibit NA activity by 50% (IC 50 ) were calculated by a 4-parameter method using MikroWin 2000 version 4 software (Mikrotek Laborsysteme GmbH, Overath, Germany).

Hemagglutination Inhibition Test
The HI test was performed to evaluate the reactivity of ferret antiserum against 2008-09 vaccine strain A/ Brisbane/59/2009, as described by the WHO manual (17). Antiserum was treated by receptor-destroying enzyme II (Denka Seiken, Tokyo, Japan) and adsorbed with packed turkey erythrocytes before testing to prevent nonspecifi c reaction. A 0.5% suspension of turkey erythrocytes was used for the HI test. Viruses with >8-fold reduced HI titer to the homologous titer of A/Brisbane/59/2009 antiserum were regarded as antigenic variants.

Statistical Analysis
To determine the cutoff value between NAI-resistant (outlier) and -sensitive viruses, box-and-whisker plots were used. The cutoff value was defi ned as upper quartile + 5.0×interquartile range from the 25th to 75th percentile. In this study, ORVs with H275Y were excluded from the overall population for statistical analysis. Outliers were excluded from the calculation of mean values and standard deviations for IC 50 .

Geographic Distribution of ORVs during the 2007-08 and 2008-09 Infl uenza Seasons
To estimate the frequency of infl uenza A (H1N1) ORVs in each prefecture of Japan, 1,734 isolates during the 2007-08 season and 1,482 isolates during the 2008-09 season were collected from all prefectures and examined by NA sequencing to detect the H275Y mutation in NA protein. In the 2007-08 season, 45 viruses possessing H275Y mutation (total frequency of ORVs 2.6%; Figure  2, panel A) were observed in 10 prefectures, indicating that the frequency of ORVs was signifi cantly lower than that in countries in Europe and the United States (8,(11)(12)(13)(14). In Tottori prefecture, however, 22 of 68 infl uenza viruses A (H1N1) tested possessed H275Y, showing a markedly higher frequency (32.4%) than that in other prefectures. In the 2008-09 season, however, ORVs were observed nationwide. Of 1,482 infl uenza viruses A (H1N1), 1,477 viruses possessed a H275Y mutation, for a total frequency of 99.7% ( Figure 2,  , and the other ORVs are also genetically close to OSVs from Japan (online Appendix Figure). These HA genes were also genetically identical or close together (online Appendix Figure), suggesting that almost all ORVs from Japan with the Hawaii lineage are derived from OSVs from Japan. On the other hand, in the Northern-Eu lineage, OSV counterparts were not observed, but foreign ORVs genetically close to ORVs from Japan were observed. During the 2007-08 season, the NA gene of ORVs from Japan was close to that of ORVs isolated from countries in Europe (i.e., A/Paris/0341/2007 and A/ England/26/2008). During the 2008-09 season, the ORVs from Japan, which shared A189T on HA protein, were further divided into 4 subclades (C-1 to C-4) by common amino acid changes on HA and/or NA (online Appendix Figure). ORVs from Japan in C-2 and C-3 were genetically close to the ORVs isolated from North America or Hawaii (e.g., A/Memphis/03/2008 and A/Hawaii/19/2008), whereas ORVs in C-1, representing most infl uenza A (H1N1) viruses from the 2008-09 season in Japan, and ORVs in C-4 were close to ORVs isolated from South Africa and Australia in the Southern Hemisphere (e.g., A/ Kenya/1432/2008 and A/Victoria/501/2008). All ORVs except C-3 were isolated before the emergence of ORVs from Japan in each subclade. These fi ndings suggest that ORVs from Japan within a Northern-Eu lineage would not have emerged domestically but instead may have been introduced from various countries.

Antiviral Drug Susceptibility
Of the 364 viruses (306 isolates in the 2007-08 season and 58 isolates in the 2008-09 season) tested by NA inhibition assay, 101 possessed a H275Y substitution. With the NA inhibition assay, although precise IC 50 values were calculated from a normal sigmoid curve (Figure 3, panels A and B), some viruses generated 2 types of unusual sigmoid curves (Figure 3, panels C and D) resulting from the mixed population of NAI-resistant and -sensitive viruses, as previously reported (18). Tentative IC 50 values were calculated from type A curves (Figure 3, panel C) and included in overall statistical analysis, but values could not be calculated from type B curves (Figure 3, panel D). Later viruses were regarded as resistant candidates.
In the NA inhibition assay for zanamivir, statistical analysis showed that 341 viruses were regarded as the zanamivir-sensitive viruses, with a mean ± SD IC 50 of 0.40 ± 0.26 nmol/L (range 0.01-1.92 nmol/L), and 16 viruses (10 ORVs and 6 OSVs) were identifi ed as outliers (cutoff IC 50 , >1.99 nmol/L) ( Table 1) (Table 1). These data suggest that D151 changes have a substantial effect on sensitivity to zanamivir (and oseltamivir). Moreover, A/Tottori/44/2008 with H275Y and D151D/G substitutions conferred high-level resistance to both NAIs ( Figure  3, panels A and B). However, a recent study reported that a D151E change was detected only after virus propagation in cell culture, but not in the original clinical specimen (19)

High Frequency of ORVs in Tottori Prefecture during the 2007-08 Season
Tottori Prefecture is located in the western part of the main island of Japan. Comprising 19 cities and geographically divided into 3 areas, this prefecture has the lowest population in Japan (Figure 4, panel B). Despite a low frequency of only 2.6% in Japan during 2007-08 season, an unexpectedly high frequency (32.4%) of ORVs was observed in Tottori prefecture (Figure 2, panel A). ORVs from Tottori were collected from 4 cities in 2 areas with no systematic bias apparent in the sampling process ( Figure  4, panel B).
Phylogenetic analysis of NA genes showed that these ORVs formed 3 subclades (Figure 4 to ORVs were observed in T-2, suggesting that ORVs in T-2 would be derived from OSVs in Tottori prefecture. A mapping study for ORVs showed that all ORVs in the Hawaii lineage were collected from Tottori city only, primarily at the end of January, whereas ORVs with the Northern-Eu lineage were collected from 4 cities, including Tottori city, during February and March. Genetically diverse ORVs belonging to T1-T3 were cocirculating only in Tottori city in the eastern area (Figure 4, panel B). The Tottori case raised concern about the possibility that these Tottori ORVs could survive to become an origin ORV for the 2008-09 season in Japan. However, phylogenetic analysis showed that all ORVs isolated during the 2008-09 season were not genetically close to ORVs from Tottori (online Appendix Figure). As a result, all ORVs from Tottori seemed to have been eliminated in the 2007-08 season, and ORVs that may have been introduced from other counties were circulating during 2008-09 in Japan.

Discussion
Our study demonstrated that ORVs dramatically increased in Japan from the 2007-08 season (2.6%) to the 2008-09 season (99.7%). All tested ORVs showed a reduction of >260-fold in susceptibility to oseltamivir by NA inhibition assay. On the other hand, almost all ORVs remained sensitive to the other antiviral-drugs, e.g., zanamivir, and M2 inhibitors. HI testing suggested that the current vaccine, A/Brisbane/59/2008, would be effective against recent ORVs. In addition, recent studies have reported that symptoms and hospitalization rates of patients infected with ORVs are no different from those seen with OSVs (14,20).  Figure 3, panels C (Type A) and D (Type B). Although the viruses with D151D/G tend to generate both patterns from repeat testing for the same samples, type B was selected in this case. §Although A/TOTTORI/44/2008 showed mixed population of D151D/G, it tended to show a normal curve fit (Figure 3, panel B). ¶The IC 50 values of most viruses with D151D/E tend to be higher than mean IC 50 values but do not exceed the cutoff value.
Japan has the largest per capita use of oseltamivir (>70%) in the world (10). Because this use could cause effi cient selection of ORVs in individual patients, Japan might be the initial site of worldwide spread of ORVs. However, long-term NAI surveillance in Japan during 1996-2007 and recent surveillance showed a low frequency of NAI-resistant viruses for any strains and subtypes (10,21,22), suggesting that transmissibility of ORVs selected by drug pressure was remarkably decreased. In addition, previous NAI surveillance (5)(6)(7)(8)(9)(10) and several animal studies (23)(24)(25)(26) also suggested that NAI-resistant viruses would become defective viruses with attenuated infectivity and transmissibility to human. In contrast, despite little NAI use, a high emergence of ORVs has been detected in several countries in Europe since November 2007. These ORVs had as effi cient transmissibility as OSVs in humanto-human transmission, resulting in worldwide spread in a short period of time. Although whether the initial ORV detected in Norway in the 2007-08 season appeared because of NAI drug pressure is unknown, those ORVs may have obtained amino acid changes on NA and/or other proteins to compensate for the defect, in addition to the H275Y substitution on the NA protein. Most ORVs belong genetically to the Northern-Eu lineage in clade 2B, suggesting that the gene constellation may contain a big advantage to retain infectivity and transmissibility.
An interesting question arose as to where the ORVs in Japan originated. In the Hawaii lineage, almost all ORVs in Japan would be derived from OSVs in Japan because the NA gene of ORVs was similar to OSV counterparts isolated at similar times or from similar regions (online Appendix Figure). On the other hand, in the Northern-Eu lineage, ORVs in Japan would have been introduced from other countries. In 2007-08, almost all ORVs would be imported from countries in Europe. In 2008-09, the ORVs in C-1, which comprised most isolates in 2008-09, and ORVs in C-4 were genetically similar to ORVs isolated from the Southern Hemisphere. Because infl uenza activity in the Southern Hemisphere occurs half a year earlier than that in the Northern Hemisphere, most ORVs in Japan conceivably could have been imported from the Southern Hemisphere. ORVs in C-2 and C-3 were genetically similar to ORVs isolated in North America and Hawaii, but the collection month of ORVs in C-3 were similar to each other, suggesting that ORVs in C-3 might be derived from an unknown common origin ORV. The ORVs obtained during 2008-09 were not genetically similar to any ORVs isolated in Tot-tori during 2007-08, indicating that ORVs from Tottori had been eliminated and had not formed the origin ORVs for the 2008-09 season in Japan. As for A/Yokohama/91/2007 belonging to clade 2C, the patient from which this virus was isolated was known to have taken oseltamivir before sampling (22), indicating that selective drug pressure in this person might have selected for this ORV.
In the NA inhibition assay for zanamivir, some viruses, including ORVs and OSVs, showed reduced sensitivity to zanamivir. NA sequencing of these viruses showed 2 types of amino acid changes. One virus, A/Tottori/16/2008 (OSV), possessed a Q136K substitution, which reportedly confers resistance to zanamivir (27,28). Conversely, most of the other viruses possessed D151 G/V/N. The amino acid changes D151 to N or E among subtype H1N1 viruses and to A, G, E, N, or V among H3N2 have been reported (7,8,19), and viruses with D151 substitutions often exhibit reduced sensitivity to NAIs (8,19,29). However, a recent study reported a possible role for cell culture in selecting these D151 variant viruses (19). In the present study, D151 variations (D151G/E/N) also were not detected from available original clinical specimens (Table 1), supporting the previous fi nding. We thus concluded that viruses with D151 variations would not have emerged naturally, and all ORVs would remain sensitive to zanamivir.
By sequencing of M2 gene, we confi rmed that almost all Japanese ORVs belonging to clade 2B retained sensitive genotype to M2 inhibitors, consistent with previously reports that recent clade 2B viruses are sensitive to M2 inhibitors, but clade 2C viruses are resistant (27).
During the 2007-09 seasons, we also addressed NAI surveillance for A/H3N2 and type B circulating in Japan and identifi ed no viruses resistant to both NAIs. Conversely, in March and early April 2009, a new swine-origin infl uenza virus A (H1N1) (now known as pandemic [H1N1] 2009 virus) emerged in Mexico and the United States and spread rapidly to many countries, including Japan (30)(31)(32)(33). In June 2009, detection of pandemic (H1N1) 2009 virus with H275Y on the NA protein was reported from Denmark, Hong Kong Special Administrative Region, People's Republic of China, and Japan, but all ORVs of pandemic (H1N1) 2009 virus emerged as sporadic cases with no evidence of effi cient human-to-human transmission (34). Although oseltamivir remains a valuable drug for treatment of pandemic (H1N1) 2009, many ORVs were isolated after prophylaxis with a half dose of the drug. Therefore, prophylaxis with oseltamivir may not be recommended as