1 Introduction

The genetic variability of influenza viruses, driven by an elevated error rate of the viral RNA-dependent RNA polymerase during replication and by genetic reassortment, favors the rapid selection of drug-resistant variants. This matter of fact has considerably limited the usefulness of the adamantane-based M2 ion channel inhibitors, amantadine and rimantadine, in the treatment of influenza A infections. Both drugs have given rise to the emergence of drug-resistant viral strains within 2–5 days of treatment in up to 30% of patients. Drug-resistant viruses are genetically stable, show unaltered virulence and transmissibility compared to sensitive viruses, and have been isolated from patients who had not been exposed to an M2 inhibitor [1].

The sialidase inhibitors zanamivir and oseltamivir were developed based on the three-dimensional structure of the enzyme–sialic acid complex. They interact with amino acids in the active site of the viral sialidase that are highly conserved in all A and B strains. It was hoped that sialidase inhibitors would not rapidly give rise to viable resistant viruses, as functionally essential residues had been targeted. This hope was initially supported by the results of clinical and animal studies, but was recently contradicted by the emergence and worldwide circulation of influenza viruses showing natural resistance to oseltamivir.

2 Mechanisms of Resistance to Sialidase Inhibitors

Because the viral HA and NA have antagonistic functions, resistance to NAI can be conferred by two types of mechanisms which are non-mutually exclusive: (a) mutations on the NA that reduce the affinity of the enzyme for the inhibitor and (b) mutations on the HA that reduce viral binding to sialic acids, and thus reduce viral dependence on NA activity. As zanamivir and oseltamivir have different chemical structures, mutations on the NA may confer resistance to one, but not necessarily to both inhibitors. In contrast, mutations on the HA usually confer resistance to both. There are a number of reports of the selection of resistant mutants upon serial passages under increasing concentrations of the NAI in Madin Darby Canine Kidney (MDCK) cells, or upon clinical treatment with the NAI. The various resistance mutations on the HA and NA are not observed with the same frequencies in vitro and in vivo.

The earlier and predominant mutations generated in vitro map to the HA gene, suggesting that the selection of HA mutants is favored on cultured cells. These mutations affect residues that are close to or belong to the receptor-binding site and reduce the affinity of HA for the sialic acids [2], whereas the affinity of the NA for zanamivir and oseltamivir remains unchanged. They allow viral growth in the presence of concentrations of NAI that inhibit the replication of a wild-type virus. In contrast to that observed in vitro, only few HA mutations have been detected in vivo (Table 1) [12, 18, 20].

Table 1 NAI resistance mutations on the NA observed in patients upon antiviral treatment
Full size table

However, the contribution of these mutations to viral NAI resistance remains to be established. It is also unclear to what extent HAs with low affinity for sialic acids expressed on human respiratory cells, such as the HAs of waterfowl influenza A viruses, can confer decreased sensitivity to NAI upon viral transmission from birds to humans. In the particular case of A/Vietnam/1203/04 (H5N1) recombinant viruses, mutations in the HA which caused a switch from avian to human receptor specificity did not alter susceptibility to the NAI in vitro [21].

The resistance mutations selected in vitro and which map to the NA gene affect conserved residues of the NA that are close to (Glu119, His274) or belong to (Arg292) the catalytic site of the enzyme (Table 2) [22, 30–35].

Table 2 NAI resistance mutations on the NA selected in vitro or studied by reverse genetics
Full size table

Despite the emergence of viruses showing resistance to zanamivir in vitro, there has been a single case of in vivo zanamivir resistance reported so far. A zanamivir-resistant type B virus with an Arg to Lys mutation at the catalytic residue 152 of the NA was isolated from an immunocompromised child treated with zanamivir (Table 1) [18]. In contrast, several types of oseltamivir-resistant viruses, harboring mutations at the Arg292 catalytic residue or at the framework residues Glu119, His274, Asn294, Asp198, or Gly402, have been isolated from oseltamivir-treated patients (Table 1) [4–10, 12–19]. The levels of resistance reported for a given mutation can vary significantly from one study to another, depending on the methodology used for IC50 determination (Table 1). In a comparative study by Mishin et al. [37], the most commonly observed R292K, E119V, and H274Y mutations were found to increase the IC50 for oseltamivir about 5,000-, 200-, and 700-fold, respectively [37]. The R292K mutation is the only one to be associated with some level of resistance to zanamivir (Table 2). In contrast, the D198N and G402S mutations, observed exclusively on type B viruses, confer cross-resistance to both oseltamivir and zanamivir, and have a moderate 5- to 10-fold effect on the IC50 (Table 1) [14, 19].

The level of resistance conferred by a given mutation is dependent on the genetic background. The R292K and E119V mutant viruses isolated in vivo were exclusively of the H3N2 subtype (Table 1). Recombinant type B viruses with these mutations were produced by reverse genetics, and showed the same resistant phenotype as the H3N2 viruses (Table 2) [27]. But attempts to rescue recombinant A/WSN/33 viruses with the R292K or E119V mutations were unsuccessful, suggesting a lethal effect in the H1N1 background [23, 24]. On the contrary, the H274Y mutant viruses isolated in vivo were exclusively of the H1N1 or H5N1 subtypes (Table 1), and recombinant H3N2 viruses with the H274Y mutation showed unaltered sensitivity to oseltamivir (Table 2) [26]. As mentioned earlier, the D198N and G402S mutations are specific to type B viruses. A notable exception is the N294S mutation, found both on a H3N2 virus and on a H5N1 virus, associated with a 300-fold resistance to oseltamivir [12], and a 15-fold resistance to oseltamivir and a 4- to 6-fold resistance to zanamivir, respectively [9] (Table 1).

The subtype specificity of resistance mutations results from distinct structural features of group 1 (N1, N4, N5, and N8) and group 2 (N2, N3, N6, N7, and N9) sialidases. Tight binding of oseltamivir carboxylate to the wild-type NA requires reorientation of the side chain of residue Glu276, which creates a hydrophobic pocket to accommodate the O-ethyl-propyl group of the inhibitor. The crystal structure of a H274Y mutant N1 in complex with oseltamivir showed that in the presence of the bulkier Tyr274 residue, the required conformational change of Glu276 is blocked and the hydrophobic pocket is not able to form, resulting in a weak binding of oseltamivir (Fig. 1a) [40].

Fig. 1
figure 1_8

Crystal structure of the active sites of a group 1 and a group 2 sialidase bound to oseltamivir carboxylate, in the absence or in the presence of a mutation conferring resistance to oseltamivir. The crystal structures of the A/Vietnam/1203/04 (H5N1) (a) and A/Tern/Australia/G70C/75 (H11N9) (b) sialidase active sites bound to oseltamivir carboxylate are represented. Carbons of the wild-type N1 (PDB #2HU4) [38] and N9 (PDB #2QWK) [39] structures are represented in green. Carbons of the H274Y mutant N1 (PDB #3CL0) [40] and R292K mutant N9 (PDB #2QWH) [39] are represented in magenta. The position of oseltamivir carboxylate (OC) in the active site of the wild-type and mutant NAs is represented in green and in magenta, respectively. In the wild-type N1 and N9 (a and b), reorientation of the side chain of residue Glu276 is stabilized by an electrostatic interaction with residue Arg224, which lets a hydrophobic pocket accommodate the O-ethyl-propyl group substituted on C6 of the inhibitor. In the H274Y mutant N1 (a), the replacement of the His274 by the bulkier Tyr274 residue blocks the reorientation of Glu276. In the R292K mutant N9 (b), the reorientation of the Glu276 side chain is prevented by an electrostatic interaction with the Lys292 residue. As a result, the hydrophobic pocket that is needed for tight OC binding is disrupted in both mutant NAs. Figures were kindly provided by Jeffrey Dyason and Mark von Itzstein

Full size image

Group 2 NAs can accommodate the H274Y substitution without disrupting the oseltamivir-binding site because they have a Thr residue at position 252 beneath residue 274, instead of the bulkier Tyr252 residue present in group 1 NAs [38, 40]. In group 2 NAs, the R292K substitution confers resistance to oseltamivir most likely due to a disruption of a hydrogen bond between the Arg292 and the carboxylate moiety of oseltamivir, and favors a Lys292-Glu276 electrostatic interaction which prevents reorientation of the side chain of residue Glu276 (Fig. 1b) [39]. In group 1 NAs, the R292K substitution is less detrimental, probably because of the presence of a conserved Tyr347 residue which makes a hydrogen bond with the carboxylate moiety of oseltamivir [38]. Such a bond is not made by the 347 residue found in group 2 NAs.

Structural data also provide clues about the absence of cross-resistance to oseltamivir and zanamivir. Zanamivir (as well as the sialic acid-based natural substrate) has a polar glycerol group at the C6 position, instead of the hydrophobic O-ethyl-propyl group characteristic of oseltamivir. Thus, binding of zanamivir does not require reorientation of Glu276, but rather involves the formation of a hydrogen bond between Glu276 and the glycerol group. As a consequence, the H274Y and R292K substitutions that block reorientation of Glu276 in group 1 and group 2 NAs, respectively, do not prevent tight binding of zanamivir [39, 40]. This chemical property of zanamivir probably contributes to the low frequency of resistance observed in treated patients. Interestingly, the structure of a N294S mutant N1 in complex with oseltamivir suggests that the loss of the Asn side chain, which points toward Tyr347 in the wild-type NA, leads to an increased flexibility of the side chain of Tyr347 and a weaker hydrogen bond with the carboxylate group of oseltamivir. The presence of a carboxylate group at the C2 position is a shared feature of oseltamivir, zanamivir, and sialic acid, consistent with the fact that the N294S mutant N1 showed weaker binding for all three compounds compared to the wild type [39, 40].

3 Impact of Resistance to Sialidase Inhibitors on Viral Fitness

Resistance mutations affect the enzymatic properties of the NA. The R292K mutation in an H3N2 background was shown to reduce the affinity of the NA for a sialic acid-like synthetic substrate 4- to 40-fold, as indicated by the increase in Km values [35, 41, 42], and to decrease its enzymatic activity (V max) [35]. Similarly, the H274Y substitution was shown to reduce the affinity of the NA of H1N1 and H5N1 viruses for a sialic acid-like substrate and/or the NA enzymatic activity [22, 24, 43]. In addition, the R292K, R152K, and E119G/V mutations may decrease the thermostability of the enzyme [27, 28, 35, 44]. Such alterations of NA properties are likely to affect the viral replicative potential.

The impact of NA resistance mutations on viral fitness was documented in both cultured cells and animal models. Recombinant H3N2 viruses carrying the mutation E119V showed the same kinetics of multiplication compared to the wild type in cultured cells [26, 28]. In contrast, H3N2 viruses with the R292K mutation, either recombinant or selected in vitro, showed a profound replication defect in MDCK or MDCK-SIAT1 cells compared to the wild type [26, 28, 35]. A H1N1 oseltamivir-resistant clinical isolate, and recombinant or in vitro-selected A/WSN/33 viruses that carry the H274Y mutation showed compromised replication efficiency compared to their wild-type counterparts on MDCK cells [23, 45, 46]. However, some R292K mutant viruses [31, 41] and a H274Y mutant A/PR/8/34 virus [24] were reported to grow as well as the wild type on MDCK cells. Overall, the in vitro replicative potential of oseltamivir-resistant viruses appears to be decreased, although to a variable extent, depending on the viral genetic background and on the cellular system used.

Mouse and ferret models were used to assess the virulence of NAI-resistant H3N2 or H1N1 isolates with the R292K, E119V, or H274Y mutations. In mice, all three mutations significantly compromised viral infectivity and/or replication ability, as indicated by the reduction in viral titers in the lungs at day 3 postinfection (1 to 2 logs for the E119V, 2 logs for the R292K, and 3 logs for the H274Y mutations) [41, 46, 47]. Reduced virulence in mice was also observed for in vitro-selected H3N2- and H1N1-resistant viruses [2, 31, 35] and, although to a lesser extent, for recombinant A/WSN/33 and A/PR/8/34 strains with the H274Y mutation [24, 48]. In the upper respiratory tract of ferrets, oseltamivir-resistant H1N1 and H3N2 isolates [41, 46, 47], as well as an oseltamivir-resistant H5N1 human isolate showing the H274Y mutation [9], replicated less efficiently than their wild-type counterparts. However, in the background of a recombinant virus derived from the A/Vietnam/1203/04 human H5N1 isolate, neither the H274Y nor the N294S mutation compromised viral pathogenicity and lethality in mice [24]. This observation suggests that viruses with a high replicative potential may accommodate mutations of resistance on the NA with no decrease in virulence.

The transmissibility of NAI-resistant viruses was examined in the ferret and the guinea pig models. A H274Y mutant H1N1 virus showed delayed transmission in a contact ferret model compared to the wild type, although a 100-fold higher dose of the mutant virus had been inoculated to donor ferrets [49]. Even more strinkingly, a H3N2 clinical isolate or a recombinant H3N2 virus carrying the R292K mutation was not transmitted in a contact ferret model under conditions that allowed a 100% efficient transmission of their wild-type counterparts [28, 50]. In contrast, recombinant H3N2 viruses with the E119V mutation were transmitted as efficiently as the wild-type virus into a contact ferret model [28] or a contact guinea pig model [29]. Overall, the transmissibility of NAI-resistant viruses seems to correlate with the degree of functional alteration of NA enzymatic properties. Noticeably, however, a E119V mutant virus transmitted only poorly when the guinea pig was used as a model for aerosol rather than contact transmission [29].

Experimental data suggested that H1N1 viruses with a H274Y mutation in the NA had such a decreased fitness and transmissibility that their diffusion in the human population was very unlikely. However, H1N1 viruses naturally resistant to oseltamivir did emerge and spread in 2007–2008. Although resistance was linked to the H274Y mutation, resistant and sensitive H1N1 viruses showed similar growth properties in vitro [51]. Enzymatic analysis suggested that a specific combination of amino acids resulted in an increased affinity of the N1 of recent H1N1 viruses for sialic acids, and that an appropriate functional balance between the activities of the HA and NA toward sialic acids was restored by the H274Y mutation on the NA [51]. These observations underline the fact that genetic variations of the hemagglutinin and sialidase, which are mainly driven by the immune response, can lead to the co-selection of resistance mutations and adventitious mutations that result in increased fitness and transmissibility.

4 Monitoring Resistance to Sialidase Inhibitors

Both phenotypic and genotypic assays can be used for the detection of NAI-resistant viruses. Viruses with resistance mutations in the NA, but not in the HA, are readily detected by using an in vitro enzyme inhibition assay to determine the IC50 (concentration of inhibitor resulting in a 50% reduction in the sialidase activity, or Inhibiting Concentration 50). Cell culture assays, which allow the determination of EC50 values (concentration of inhibitor resulting in a 50% reduction in the viral yield, or Effective Concentration 50), account for both NA- and HA-mediated resistance. Both assays are thus useful and complementary.

The NA inhibition assay is based on a fluorometric NA activity assay [52] which was further optimized for IC50 determination [53]. The fluorescence of the 4-methylumbelliferone released upon NA-mediated cleavage of the fluorogenic substrate 2′-(4-methylumbelliferyl)-α-d-N-acetylneuraminic acid (MUNANA) is measured in the absence or in the presence of antiviral drugs. A more sensitive assay was developed, using a chemiluminescent substrate (1,2-dioxetane derivative of sialic acid, NA-STAR), [54] instead of the MUNANA. The IC50 values show some variability from one study to another, depending on whether the fluorogenic or chemiluminescent substrate is used and on the substrate concentration that is used [55]. Both substrates have also been used to determine the affinity of the NA for zanamivir and oseltamivir (Ki) [22, 41, 43, 56].

The evaluation of viral sensitivity to NAI in cell culture assays is usually based on the degree of inhibition of virus yield upon multicycle growth (as measured by hemagglutination or pfu titers, or by TCID50), viral plaque formation (as determined by the number and size of plaque-forming units), or viral infectivity (as determined by a FACS assay following immunohistochemical staining with an anti-NP antibody) [57]. Such assays were initially developed in the standard MDCK cell line. A poor correlation was found between the sensitivity of clinical isolates to NAIs in cell culture and their sensitivity in the enzyme inhibition assay [58, 59]. The fact that sensitive isolates generally behaved as drug resistant in cell culture assays was probably due to the low amounts of sialyl-α2,6-galactose-containing receptors on the surface of MDCK cells, which reduced viral dependency on NA sialidase activity. Conversely, a drug-resistant isolate with the R152K mutation appeared as drug sensitive on MDCK cells because of the presence of a mutation in the HA, which increased viral binding to sialyl-α2,3-galactose-containing receptors on the surface of MDCK cells, and thus increased viral dependency on NA sialidase activity [18]. Assays performed on MDCK cells overexpressing the human α2,6-sialyltransferase (MDCK-SIAT1 cells) proved to be a more suitable system to monitor sensitivity of human viruses to NAIs [60, 61]. Primary normal human bronchial epithelial cells, which are functionally and morphologically close to normal differentiated human airway epithelium, may provide an optimal system for prediction of virus resistance in vivo [21, 62].

Resistance mutations present in the NA gene are classically detected by viral RNA extraction, RT-PCR, and sequencing of the resulting amplicon. In addition to this accurate but time-consuming method, novel approaches are being developed for the rapid detection of well-characterized resistance mutations. For instance, a one-step allelic discrimination RT-PCR assay using two distinctly labeled probes specific for the wild-type His274 and the mutant Tyr274 codon allowed the rapid detection of the H274Y substitution in circulating H1N1 viruses [63] and in highly pathogenic avian influenza H5N1 [64]. Pyrosequencing protocols were recently developed for rapid detection of the most commonly reported mutations associated with NAI resistance in seasonal H1N1 and H3N2, and in H5N1 viruses [65–68]. Interestingly, pyrosequencing enabled accurate quantification of the relative proportions of resistant cultured viruses when experimentally mixed with sensitive cultured viruses and when present in the total virus population at levels as low as 10% [65, 67, 68].

5 Incidence of Resistance to Sialidase Inhibitors Under Treatment

In immunocompetent patients, the emergence of seasonal influenza viruses resistant to NAIs in clinical trials was observed at much lower frequency compared to the emergence of amantadine-resistant viruses [1, 69]. No zanamivir-resistant virus has been isolated from immunocompetent patients treated with zanamivir so far, which could be partly due to the infrequent usage of zanamivir in the clinic and in clinical trials compared to oseltamivir. Viruses with a decreased sensitivity to oseltamivir associated with a H274Y mutation were detected in up to 4% of adult volunteers who had been experimentally infected with H1N1 viruses and subsequently treated with oseltamivir [3]. Based on cumulative data from Roche-sponsored clinical trials involving more than 1,000 adults and 500 children, the incidence of viruses with reduced susceptibility to oseltamivir is about 0.32–0.4% in adults and 4–5% in children infected with influenza A viruses [70].

The higher frequency of emergence of oseltamivir-resistant viruses in children could be related to a higher viral replication, a prolonged viral shedding, and an increased clearance of the drug compared to that in adults [71]. These findings led to the recommendation of a weight-dependent dosage of oseltamivir in children [72]. Two Japanese studies reported the isolation of drug-resistant viruses in 5.5% and 16–18% of children treated with the initial 2 mg/kg dosage following infection with H1N1 and H3N2 viruses, respectively. A reduced incidence rate of 1.1% was reported in recent studies when the weight-dependent dosage was used [70]. In two independent studies, the incidence of resistant influenza B viruses in children treated with oseltamivir was found to be <1% [4, 19]. Notably, however, influenza B viruses are naturally less sensitive to oseltamivir compared with influenza A viruses (Table 3), and the current oseltamivir dosage may be suboptimal with regard to the emergence of resistant influenza B viruses.

Table 3 IC50s to zanamivir and oseltamivir for seasonal influenza A and B viruses
Full size table

In immunocompromised patients under treatment with NAIs, the emergence of viruses showing reduced drug susceptibility has been reported in at least nine cases (Table 4). The only reported case of emergence of a zanamivir-resistant virus upon treatment occurred in an immunocompromised child infected with influenza B virus (Table 4).

Table 4 NAI-resistant viruses isolated from immunocompromised patients
Full size table

Persistent viral shedding during 12 and 17 months was reported in an immunocompromised child [15] and in an immunocompromised adult [6], respectively. It was suggested that such a prolonged viral shedding may favor the selection of resistant viruses [83]. However, given the limited number of case reports, the incidence of viruses resistant to NAIs in immunocompromised patients remains unknown.

Emergence of oseltamivir-resistant viruses during treatment was recently described in three patients infected with a highly pathogenic H5N1 virus [9, 10]. When sensitivity to oseltamivir was systematically monitored in eight patients during treatment, the emergence of a resistant H5N1 virus was observed in two of them (25%) [10]. These observations suggest that the incidence of resistant virus could be higher upon infection with H5N1 viruses compared to seasonal viruses. The difference could be due to higher viral loads and to a prolonged viral shedding due to the lack of previous immunity [84].

6 Incidence of Resistance to Sialidase Inhibitors in the Absence of Treatment

The emergence of resistant viruses that are able to spread in the human population is a public health concern, as it compromises the effectiveness of the drug. The dramatic increase in the proportion of H3N2 viruses naturally resistant to amantadine worldwide from 1–2% in 2002 to 90.5% in 2005–2006 [69] stressed the need for a global surveillance of viruses resistant to NAIs, and in particular to oseltamivir as the drug became widely used in some countries.

Notably, there is no widely accepted definition of resistance to NAIs. A unique IC50 threshold value cannot be easily defined, because IC50 values may vary depending on the experimental conditions used for determination, and may vary from one type/subtype of influenza virus to another when determined in parallel. A greater sensitivity to oseltamivir compared to zanamivir was generally observed for viruses of the H3N2 subtype, whereas the opposite trend was observed for H1N1 viruses (Table 3). Influenza B viruses are generally found to be less susceptible to NAIs than influenza A viruses, as indicated by 2–10-fold and 5–50-fold increased IC50 values for zanamivir and oseltamivir, respectively (Table 3). In surveillance studies, statistical analysis of the data obtained on a large panel of viruses has been used to define the mean IC50 of sensitive viruses within a given NA type/subtype [7, 73, 76, 80, 85]. For example, the worldwide survey performed by the Global Neuraminidase Inhibitor Susceptibility Network from 1999 to 2002 found that the mean IC50 values for oseltamivir and zanamivir were 0.37 and 1.38 nM, respectively, for H3N2 viruses (n = 1,119), 0.51 and 0.40 nM, respectively, for H1N1 viruses (n = 622), and 1.86 and 5.37 nM, respectively, for B viruses (n = 743) (Table 3) [85]. Viruses are defined as outliers if they show an IC50 value higher than the mean IC50 + 2SD or the mean IC50 + 3SD, depending on the study.

Among the outliers group, viruses with an IC50 value at least 10-fold higher compared to the mean IC50 of sensitive viruses are considered as resistant viruses if their NA gene contains a mutation already known to be associated with drug resistance; if not, they are considered as extreme outliers [7, 73, 77]. The NA mutations found in extreme outliers are of interest since most of them affect conserved residues located close to the catalytic site, e.g., residue 222 for type B viruses [19, 77] or residue 151 for H3N2 viruses [7, 73]. Even more striking, a type B virus isolated in the absence of treatment showed the R371K mutation [7]. Residue 371 is part of the Arg triad (Arg118, Arg292, and Arg371) shared by all influenza NAs together with bacterial sialidases [86]. The R371K mutant type B virus showed a stronger decrease in sensitivity to zanamivir and oseltamivir (40- and 500-fold, respectively [7]) than some viruses defined as truly resistant. This observation underlines how difficult it is to figure out the boundary between natural variations in the susceptibility to NAIs and true resistance.

Two worldwide surveillance studies performed through the 1996–1999 and 2004–2007 periods reported 3/1,054 (0.28%) and 10/3,253 (0.3%) extreme outliers, respectively [7, 73], suggesting that the frequency of extreme outliers remains stable. In contrast, the number of oseltamivir-resistant viruses isolated from untreated patients increased in the recent years. Noteworthy, H1N1 viruses with the H274Y mutation (7 in total) and B viruses with the D198N mutation (3 in total) were the only kind of oseltamivir-resistant viruses isolated from untreated patients from 1999 to 2007, suggesting that these two mutations are the less costly in terms of viral fitness [7, 19, 80, 85].

During the 2007–2008 influenza season, H1N1 viruses naturally resistant to oseltamivir emerged and spread in the human population. Resistance was conferred by the H274Y mutation of the NA. Resistant viruses were more prevalent in Europe (25%) than in the Americas (16%) or the Western Pacific region (5%) (Fig. 2) (http://www.who.int/csr/disease/influenza/ResistanceTable200806013.pdf).

Fig. 2
figure 2_8

Worldwide prevalence of oseltamivir-resistant H1N1 viruses, as of 01 July 2008. A color code is used to represent the percentage of oseltamivir-resistant H1N1 viruses. Yellow: < 1.0%; light orange: 1.0–9.9%; dark orange: 10.0–24.9%; red: > 25%; gray: data not available. Data source: WHO/GIP, Map Production Public Health Information and Geographic Information Systems (GIS), http://www.who.int/csr/disease/influenza/Global_H5N1Resistance_20080701.jpg With authorization

Full size image

The highest prevalence was observed in Norway where oseltamivir usage is moderate (67%) [87], whereas in Japan where oseltamivir is used most frequently, the prevalence was low (3%) [88]. Although the prevalence of H1N1-resistant viruses was not correlated with oseltamivir usage, their emergence may be related to selective drug pressure. According to WHO [http://www.who.int/csr/disease/influenza/h1n1_table/en/index.html (accessed 23 March 2009)] and to recently published data [89], H1N1 viruses isolated in 2008–2009 were still predominantly resistant to oseltamivir.

7 Clinical Consequences of Resistance to Sialidase Inhibitors

In immunocompetent patients, the emergence of oseltamivir-resistant variants does not seem to impact the evolution of influenza syndrome. Cohort studies indicate that the intensity of symptoms, the duration of symptoms, and the course of viral excretion were not modified by the emergence of resistant viruses [4, 8, 12, 19]. A recent study, based on 265 cases of infections with H1N1 viruses sensitive or naturally resistant to oseltamivir in 2007–2008, reports no differences in primary symptoms, viral shedding, overall complication, and hospitalization rates. However, an insignificant trend of increased evolution to pneumonia and sinusitis in patients infected with resistant viruses was observed [87]. These overall conclusions are drawn from a limited number of cases and should be confirmed in the future.

In immunocompromised patients, viral respiratory tract infections have a significant impact in terms of morbidity and mortality. A retrospective study including more than 3,500 adult cases found an incidence of 0.3–0.4% in kidney- or liver-transplanted patients and about 4% in lung-transplanted patients [90], but the clinical impact of these infections remains to be explored. In bone marrow-transplanted patients, the incidence of viral respiratory tract infections is about 1%, and influenza viruses are isolated in up to 47% cases. In these patients, influenza viruses cause severe viral pneumonia, and mortality rates in the range of 4.5–47% depending on the studies [91]. An antiviral treatment is needed in order to achieve efficient viral clearance, and the emergence of antiviral resistance is of major concern. Seven cases of emergence of NAI-resistant influenza viruses associated with prolonged viral shedding in immunocompromised patients have been described so far, five of which were fatal (Table 4) [6, 14, 15, 18].

In patients infected with highly pathogenic H5N1 viruses, as in immunocompromised patients, lack of antivirus-specific immunity and high levels of viral replication may contribute to the severity of the disease and favor the development of resistance. In H5N1 virus-infected patients treated with oseltamivir, the emergence of a drug-resistant virus carrying the H274Y mutation was associated with fatal outcome in two of three cases [9, 10]. In one of the fatal cases, oseltamivir was administered early in the course of infection and the clinical condition remained stable during the first three days of treatment. The later isolation of an oseltamivir-resistant virus coincided with clinical deterioration and increased viral loads, suggesting that the development of drug resistance was responsible for treatment failure [10].

Overall, these observations highlight the need for additional antiviral agents that could be used instead of or in combination with oseltamivir for the treatment of H5N1 infections and for the treatment of immunocompromised patients.

8 Conclusion

Sialidase inhibitors are currently the drug of choice for the treatment of seasonal influenza, and their stockpiling is a key element of the pandemic preparedness plans in several countries. The development of viral resistance to NAIs is of clinical concern on an individual level in immunocompromised patients and in patients with zoonotic influenza. It becomes of concern at a population level as soon as resistant viruses are efficiently transmitted from human to human and are able to spread, because the clinical effectiveness of the drug is compromised overall. The emergence and circulation of influenza H1N1 viruses showing natural resistance to oseltamivir during the 2007–2008 season were an unexpected event, because clinical trials had resulted in low frequencies of oseltamivir resistance in adults, and animal studies had pointed to a negative impact of resistance mutation on viral fitness and transmissibility. The genetic and enzymatic characterization of these recent H1N1 viruses suggests that drift mutations on the NA contribute to the absence of debilitating effect of the H274Y resistance mutation. These findings stress the need for a better understanding of the interplay between resistance mutations and drift and/or compensatory mutations, on the NA, the HA, or possibly on other genomic segments, for all subtypes of circulating viruses. The impact of genetic drift on natural susceptibility of influenza viruses to NAIs, in the absence of a resistance mutation, also needs to be explored through further clinical, in vitro, and animal studies.

Strategies for antiviral resistance containment and surveillance still have to be improved. Enhanced methods for the rapid and accurate detection of resistant viruses are requested, as well as increased coverage and testing capacity of surveillance networks, and reinforced vigilance in the therapeutic usage of NAIs. The further development of novel anti-influenza drugs, targeting either the NA or other viral proteins such as the polymerase or the NS1 protein, is essential. Such developments would offer an alternative if circulating viruses become predominantly resistant to currently available compounds. They would also open the way to combination therapies, which are likely to reduce the emergence of drug-resistant viruses in at-risk populations such as children, immunocompromised, and H5N1 patients.