Antiviral Research 49 (2001) 147–156 www.elsevier.com/locate/antiviral

Review Position statement: global neuraminidase inhibitor susceptibility network

Maria Zambon a,1, Frederick G. Hayden b,*,1 (On behalf of the Global Neuraminidase Inhibitor Susceptibility Network)

a Public Health Laboratory Ser6ice, London, UK b Uni6ersity of Virginia Health Sciences Center, Box 800473, Department of Internal Medicine, Charlottes6ille, VA 22908, USA Received 21 December 2000; accepted 11 January 2001

1. Introduction of drug resistance in influenza viruses following the introduction of the influenza neuraminidase The global Neuraminidase Inhibitor Suscepti- inhibitor (NI) class of antiviral agents. The first bility Network (NISN) was established in 1999 to meeting was held on December 13, 1999, and address public health and regulatory concerns re- subsequent meetings were conducted on June 30, garding the potential emergence and consequences 2000 and September 23, 2000. The Network ini- tially included investigators and public health per- * Corresponding author. Tel.: +1-804-9245059; fax: +1- sons with demonstrated interest in neuraminidase 804-9249065. inhibitors or anti-viral resistance and has ex- E-mail address: [email protected] (F. G. Hayden). panded to include representatives of each of the 1 Other current Network members: Michele Aymard, Uni- four WHO global influenza reference laboratories versite Claude Bernard, Lyon, France; Alan Hay, National and scientists from regions of the world where Institute for Medical Research, Mill Hill, London, UK; Alan Hampson, WHO Collaborating Centre for Reference and increasing use of these drugs is anticipated. The Research on Influenza, Melbourne, Australia; Alexander Network’s activities are funded currently by two Klimov, Centers for Disease Control, Atlanta, GA; Jenny corporate sponsors, GlaxoSmithKline and Hoff- McKimm-Breschkin, Biomolecular Research Institute, Mel- man-LaRoche, and company representatives are bourne, Australia; Arnold Monto, University of Michigan, invited to attend meetings as observers. However, Ann Arbor, MI; Masato Tashiro, WHO Collaborating Center for Influenza, Tokyo, Japan; Robert Webster, St. Jude Chil- the deliberations and actions of the Network itself dren’s Research Hospital, Memphis, TN. Daniel Lavanchy, are intended to be independent of any company, Coordinator, Epidemic Disease Control, Department of Com- and the core working group of the Network is municable Disease Surveillance and Response, WHO, Geneva, composed of scientists drawn either from aca- Switzerland is an advisor to the Network. Pharmaceutical sponsor observers include Margaret Tisdale, Rob Fenton, and demic or public health sectors. The broad objec- Michael Elliot (GlaxoWellcome) and Noel Roberts, Emma tives of the Network are to: (1) provide a coherent Covington, and Penelope Ward (Hoffman-LaRoche). approach to global NI resistance monitoring from

0166-3542/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S0166-3542(01)00124-3 148 M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156 both public health and research perspectives; (2) are rooted in the accumulated experiences with the examine data from the scientific literature and from first class of anti-influenza agents, the M2 protein specific monitoring programs to make recommen- inhibitors amantadine and rimantadine, resistance dations for appropriate general strategies and spe- to drugs for HIV infection, and on the significant cific assays for monitoring resistance; (3) conduct problems associated with the emergence of antimi- longitudinal prospective surveillance for resistance crobial resistance in non-viral pathogens (pyogenic emergence through a link with the existing WHO bacteria, tuberculosis, malaria). Indeed, the naivety Global Influenza Surveillance Network; and (4) of the mid 20th century towards the conquering of communicate this information to the scientific infectious diseases has been replaced by mature community. In particular, the Network will select realism and respect for the mutability of microbial appropriate monitoring assays, determine the NI pathogens. susceptibility of representative clinical isolates (\ The experience with the M2 inhibitors, aman- 1000) collected before introduction of these drugs, tadine and rimantadine, illustrates the potential of and continue surveillance for resistance emergence influenza A viruses to rapidly develop drug resis- on a sustained basis (\5 years). The four WHO tance in the clinical setting (reviewed in Hayden, Collaborating Centers for Reference and Research 1996). The basis of resistance is point mutations in on Influenza will continue to provide the Network the M gene with corresponding single amino acid with viruses isolated in the post-licensing period. changes in the M2 protein that confer high level The current position paper of the Network covers cross-resistance between the drugs in vitro (Hay, the rationale, mechanisms of NI resistance includ- 1992; Belshe et al., 1988). Rarely, amantadine-resis- tant variants predominate in clinical isolates. How- ing phenotypic and genotypic characterization, ever, resistant variants are present in low currently recommended approaches and assays, concentration in virus samples and emerge within and future directions for neuraminidase inhibitor 2 to 5 days of initiating drug therapy in 30% of resistance monitoring in influenza viruses. This treated immunocompetent adults and children consensus statement derives from deliberations of (Belshe et al., 1988; Hayden et al., 1991; Hall et al., the group over the past 18 months and is intended 1987). More prolonged virus replication, as seen in to provide the foundation for subsequent commu- immunocompromised hosts, is associated with high nications which will deal with assay selection and frequencies of resistance emergence (Englund et al., validation, statistical considerations, baseline sus- 1998). Amantadine-resistant variants are geneti- ceptibility results, and surveillance data. cally stable, are not reduced in infectivity or viru- lence in animal models, cause typical influenza illness in humans, and are transmissible from 2. Antimicrobial resistance person to person under conditions of close contact. Spread of these resistant variants has caused fail- The emergence and societal implications of an- ures of drug prophylaxis in households and nursing timicrobial resistance has received increasing atten- homes (Hayden et al., 1989; Degelau et al., 1992; tion in recent years. The development of effective Mast et al., 1991). Thus, amantadine/rimantadine- antiviral drugs is an important scientific achieve- resistant variants possess the biological properties ment and has led to the licensing and use of over associated with clinically, and possibly epidemio- two dozen specific antiviral drugs in the developed logically, important drug resistance. To date the use world in the last two decades. The most recently of amantadine and rimantadine has been relatively licensed class of antiviral compounds, the neu- limited and has been associated uncommonly (B raminidase inhibitors of influenza viruses 1%) with recovery of resistant variants in the (zanamivir [Relenza™, GlaxoSmithKline] and os- general population (Ziegler et al., 1999). eltamivir [Tamiflu™, Roche]) has sparked opti- In part because of this experience, the introduc- mism but also controversy. Inevitably, the licensing tion of the neuraminidase inhibitors (NI) requires of new antiviral drugs provokes concerns about the the establishment of longitudinal surveillance to development of antiviral resistance. Such concerns determine the possible emergence, clinical impor- M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156 149 tance, and epidemiological consequences of resis- technical problems which limit the usefulness of tant strains, particularly in patient populations many current assays (e.g. cell culture-based phe- under-represented in clinical trials. This necessity notypic assays). may be translated into regulatory requirements Clearly, there is a need to determine the poten- following the introduction of new antiviral drugs. tial for development of resistance to NIs during Although there has been limitation on public sec- wide spread use for the treatment of acute influ- tor prescription of NIs in several European coun- enza, and the possible consequences should resis- tries because of reimbursement issues (NICE, tance develop. Key questions include the 1999), there was extensive prescribing in primary frequency and rapidity of resistance development; care in the USA and to a lesser extent in western the genetic stability, virulence and transmissibility Europe during the 1999–2000 influenza season. of any resistant strains; and whether alterations in Thus the possibility of emergence of antiviral drug susceptibility result in antigenic changes in resistance is realised for the first time as the drugs circulating strains. Answers to these questions are used outside of the clinical trials setting. With may be technically challenging to achieve. Assess- the anticipated extension of NI availability to ment of NI resistance should therefore include: other countries and increased awareness of their “ Use of suitable, validated assays for resistance value, it is expected that use will continue to testing with appropriate controls. Currently, increase, and with it the potential for drug pres- measuring inhibition of NA enzymatic activity sure to select resistant variants. in vitro is the most sensitive and specific phe- notypic means of detecting NA variants, due to the lack of predictive cell culture-based assays. 3. Requirements for surveillance of NI resistance “ A suitable range of viral isolates to establish baseline susceptibility prior to the introduction Influenza is a global disease with seasonal vari- of NI drugs. ability and geographic unpredictability. The emer- “ Analysis of post treatment isolates, particularly gence and spread of new variants is rapid and from populations at higher risk for protracted relentless. Comprehensive NI resistance surveil- virus replication (e.g. infants and young chil- lance therefore needs to extend globally. One im- dren, immunocompromised hosts, elderly portant advantage in monitoring the emergence of adults). NI resistance is that the WHO has a well-estab- “ A panel of well-characterized resistant viruses lished surveillance network for monitoring the to incorporate into the screening assays. impact of influenza worldwide and the associated “ All marketed NIs. antigenic and genetic changes of the responsible viruses (WHO, 1996) (www.who.int/health-topics/ influenza.html). Monitoring for antiviral suscepti- 4. Mechanism of action of NIs and basis of bility is important not only in its own right, but resistance also for the potential impact antiviral drugs may have on the generation of antigenic diversity. NA and haemagglutinin (HA) work in concert Comprehensive surveillance of NI resistance re- during viral entry and release from the cell. At cell quires co-operation of national/regional govern- entry, HA binds to the cell via receptors bearing mental agencies and other public health terminal sialic acid residues. Following budding, authorities responsible for influenza surveillance, progeny viruses remain attached to the host cell as well as liaisons with the pharmaceutical groups and to each other through HA binding to sialic responsible for antiviral drug development and acid-bearing receptors on the cell surface and on marketing. In addition, the generation of mean- the HA and NA of progeny virus. The viral ingful data on NI resistance, wherever it is carried neuraminidase (NA), which is required to com- out, requires robust, reproducible assays of drug plete the viral replication cycle, cleaves the sialic susceptibility and an understanding of the specific acid residues from these receptors and facilitates 150 M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156 release of new virions from infected cells and 5. Resistance in vivo spread of virus within the respiratory tract (re- viewed in Gubareva et al., 2000). The structure of Neither NA nor HA mutations have been se- the active site is highly conserved across all nine lected readily in vitro. Sequential passage in vitro influenza A NA subtypes and influenza B (re- has typically led to appearance of HA variants viewed in Colman, 1994). The essential role of NA first and NA mutants only later (Tisdale, 2000). in viral replication and the conservation of the The NA mutants selected in vitro usually but not active site thus make the NA an attractive target necessarily predict those observed in vivo. Pre- for drug action. The NIs bind to the catalytic site clinical resistance studies in animals (Mendel and of the NA and competitively inhibit this key viral Sidwell, 1998; Sidwell et al., 1998) and monitoring function. of influenza isolates during clinical trials with Resistance to NIs has been shown to arise in zanamivir (Boivin et al., 2000; Barnett et al., 2000) vitro and in vivo by two mechanisms to date and oseltamivir (Treanor et al., 2000), although (reviewed in McKimm-Breschkin, 2000 and in limited in number, suggest that resistance will not Mendel and Sidwell, 1998): develop rapidly. To date the frequency of recovery “ Mutations in HA which decrease virus recep- of resistant virus with NA mutations was 1–2% tor binding affinity. This facilitates progeny in immunocompetent adults receiving oral os- virus release and reduces the requirement for eltamivir for acute treatment of influenza and has NA enzyme activity, hence reducing viral sen- not been observed in immunocompetent persons sitivity to any NIs. However, such mutations receiving inhaled zanamivir. Furthermore, resis- in HA can also decrease the infectivity of the tant isolates studied have been compromised in virus by reducing binding at viral entry. It is their NA activity or stability, and in the majority not known currently if HA mutations alone of cases this has translated into a reduction in can produce clinically important NI resistance virus infectivity/replicative ability (Table 1) (re- in vivo. viewed in Tisdale, 2000; McKimm-Breschkin, “ Mutations in NA which decrease inhibitor 2000). Most of the oseltamivir-resistant clinical binding affinity. Such mutations have also isolates possess a mutation at Arg292Lys. More- caused reduced enzyme catalytic activity or over, it will be necessary to consider resistance to stability and have often resulted in reduced each neuraminidase subtype separately, even replication and decreased virulence in vivo though the active site is highly conserved across all (Table 1). subtypes. For example, oseltamivir has been

Table 1 NA mutations observed in preclinical and clinical studies and their effects on enzyme functiona

InhibitorNA NA mutations Selected Enzyme function Type/subtype

In vitro In clinic

ZanamivirA/N2, B E119G/A/D Yes No Reduced stability A/N9 R292K Yes No Reduced catalytic activity (B20% wild-type) B R152K No Yes Reduced catalytic activity (3–5% wild-type) OseltamivirA/N2 R292K Yes Yes Reduced catalytic activity (B20% wild-type) A/N2,N9E119V No Yes Reduced catalytic activity A/N1H274Y Yes Yes Reduced catalytic activity

a N2 numbering is used for all types and subtypes (Colman, 1994). M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156 151 shown to select for Arg292Lys in N2 and for nately a(2,3) linkages. If HA binds with lower His274Tyr in N1 containing influenza A viruses affinity to receptors, then the virus is less dependent both in vitro and in treated persons. Large-scale on NA activity for release and appears less sensitive studies are required to monitor this further during to NIs in vitro. Furthermore, many low passage widespread use of drugs in different countries. clinical isolates plaque so poorly that such assays are not feasible. In addition, at least one influenza B isolate recovered from an immunocompromised child receiving nebulised zanamivir was fully sensi- 6. In vitro assessment of resistance tive in MDCK cells but resistant in a NA enzyme inhibition assay (Gubareva et al., 1998). This There is no clear relationship between the pheno- variant has a NA catalytic site mutation that types and genotypes of viruses that emerge during confers resistance to NIs but also an HA mutation exposure to NI drugs in vitro, and this reflects the that results in enhanced binding to MDCK cell complexity of resistance to NI and the need to receptors and apparently increased susceptibility to consider both the NA and HA virus components zanamivir in this cell type. (Blick et al., 1998; Barnett et al., 1999). Although useful for evaluating resistant isolates selected in 6.1. HA-related resistance cell culture, the standard methods for detecting viral drug resistance based on changes in antiviral If HA binds less tightly to host receptors, the phenotype in cell culture (plaque reduction assay, virus elutes more easily from receptors and is less yield reduction, EIA) have, so far, not proved dependent upon NA activity for release of progeny reliable for screening virus isolates from clinical virus. Hence the virus will appear less sensitive to trials of NIs. Both false positive and false negative NIs in cell culture-based assays. Mutations in HA resistance results have been recognized (Gubareva which change receptor binding may mask resis- et al., 1998; Penn et al., 1996; Gubareva et al., tance due to NA mutations in cell culture. Weak 2001). For example, zanamivir susceptibility in vivo HA binding and/or disproportionately high NA in experimental animal models of influenza corre- activity may result in failure of virus to infect cells lates well with in vitro susceptibility determined by due to release of attached virus before viral cell NA inhibition assay but not with plaque assay in penetration occurs. Such a phenomenon has been MDCK cells (Tisdale, 2000). described in in vitro-selected, drug dependent HA Limited passage clinical isolates tend to give mutants (Barnett et al., 1999). Conversely, a higher antiviral IC50 values in MDCK cells (i.e. strongly binding HA could mask a resistant NA. they appear much less sensitive to NI) compared HA mutations resulting in altered sialic acid with laboratory strains of virus. Such viruses usu- receptor binding would be expected to occur in ally have a fully susceptible NA by enzyme inhibi- the sialic acid receptor region of HA1, although tion assay and are inhibited in growth in human other sites have been reported from in vitro stud- cells (explant cultures) and in animals (Penn et al., ies (reviewed in McKimm-Breschkin, 2000). How- 1996; Woods et al., 1993). For example, the MDCK ever, mutations which result in decreased affinity plaque assay IC50 values of representative sets of of HA to one cell receptor type may not confer clinical isolates ranged over 700-fold, whereas vari- decreased affinity for receptors on other cell types. ation in an NA inhibition assay values was gener- Other carbohydrate residues on the cellular recep- ally within a 10-fold range (Woods et al., 1993). tor and glycosylation of the HA may also affect Such results probably reflect the sub-optimal bind- receptor binding. Hence, mutations observed in ing of clinical isolates to the a(2,3)-linked sialic HA during in vitro studies to select NI-resistant receptors of MDCK cells. Human influenza viruses virus may be (relatively) specific for the cell type bind preferentially to sialic acid linked to the used in the experiment (usually MDCK) and may penultimate galactose by an a(2,6) linkage, whereas not be predictive of HA mutations required to MDCK and many other cell types have predomi- reduce binding affinity to receptors in the human 152 M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156 respiratory tract (Gubareva et al., 2000). Con- (\10-fold) than the limit of variability noted in versely, mutations in HA which decrease affinity pre treatment isolates. The absolute concentration for the receptors in the human respiratory tract above which resistance may be predictive of clini- may not have decreased affinity for receptors in cal failure (i.e. lack of antiviral effects in vivo) laboratory cell lines. A cell line carrying a(2,6) also has still to be determined, but will probably linked sialic acid receptors reflective of the human be greater than 50 nM. An assessment of the respiratory tract and which supports growth of potential to lose mutants during cell culture or to fresh clinical isolates is not currently available. fail to detect the activity of mutant NAs is re- Phenotypic assays for HA mutations (e.g. RBC quired. To confirm resistance, the virus NA elution, binding to artificial receptors) are not yet should also be sequenced to identify the genotypic standardized and require further development. change(s) responsible for the change in pheno- These observations present a major, and as yet type. Optimally such sequencing studies should be unresolved problem for phenotypic assay of HA- performed on original clinical samples to exclude mediated resistance to NI drugs. Further, expan- ex vivo selection of resistance mutations during sion of virus from clinical samples may allow growth of virus in cell culture systems containing variants in HA to arise as the virus adapts to the residual drug. cell type used for culture. Currently, comparison of the sequence of HA from the pre- and post- treatment samples, looking specifically for muta- 6.3. Determination of endpoints tions at or near residues involved in sialic acid binding, may be the best option to evaluate emer- For the NI Susceptibility Network there is a gence of mutations in HA. The occurrence of need to determine as accurately as possible the natural (i.e. not drug induced) variants in HA on susceptibility of a large number of clinical isolates passage through a patient must be expected and throughout the world. This should allow detection factored into the outcomes analysis. of significant changes in susceptibility occurring in the circulating viruses from year to year. Highly 6.2. NA-related resistance accurate statistical analysis may prove useful in identifying resistant variants should they arise and Since NA functions extracellularly and NIs are will be dependent on precise in vitro measures to active without entering the host cell, direct NA define the usual ranges of susceptibility. inhibition assays are likely to be more predictive The 50% inhibitory concentration (IC50)isthe of in vivo resistance than cell culture-based as- most precise value for this purpose (Richman, says. Decreased sensitivity to NIs due to muta- 1996). The relationship between the inhibition of tions in NA may be assayed using in an in vitro NA activity and the log of the drug concentration enzyme assay using an artificial substrate. Al- is usually a sigmoid curve and the IC50 is derived though assays have been developed to detect infl- from the center of the linear portion of the curve. uenza NA enzymatic activity directly in clinical Other endpoints (e.g. IC90 or IC99) may permit the samples (i.e. Zstat® influenza test), usually virus detection of a heterogeneous mixture with resis- must be expanded in cell culture prior to suscepti- tant sub-populations, but their calculation is bility assay. Decrease in sensitivity of the NA to much less precise and subject to artifactual errors.

NI between pre and post treatment samples may The IC50 determination of susceptibility is not a indicate resistance. However, the high level po- therapeutic target concentration, which must be tency of NIs means that clinical isolates may have determined independently for each drug (Rich- a wide range of sensitivity in an NA assay but man, 1996), and may not detect low level resis- remain inhibited by clinically achievable concen- tance or resistant subpopulations. Inspection of trations of the drug. Shifts in sensitivity which the inhibition curves is useful for the latter, partic- reliably predict clinical resistance need to be ularly when there is failure to inhibit NA enzyme defined, but will probably be substantially higher activity fully at high NI concentrations. M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156 153

While it may be valuable to relate achievable tion. When sequential respiratory samples are drug concentrations to antiviral activity, this is taken and shown to progress from culture positive not necessarily an accurate predictor of clinical to culture negative, it may reasonably be assumed efficacy and is dependent on various pharmacoki- that a person has cleared the virus. Provided the netic parameters that will be unique to each drug. last culture positive sample is not resistant, it is

Also the relationship of the IC50 to the IC90 will highly unlikely that subsequent culture negative vary with different inhibitors and will be depen- samples from the same patient will contain clini- dent on the slope of the linear part of the curve. cally important levels of resistant virus. It there- To understand the level of susceptibility that is fore follows that patients whose samples are clinically relevant for each inhibitor will require culture positive pre-treatment but culture negative clinical efficacy data and/or additional analysis in at all post-treatment sample times do not carry a suitable in vivo model. clinically important levels of drug resistant virus.

The observed IC50 value is also influenced heav- In calculating the incidence of resistance, patients ily by the type and concentration of the substrate, whose post-treatment virus samples were resis- as reflected by the following equation: IC50 = tance assay negative or culture negative can rea- Ki ×(1+[S]/Km) where Ki, binding constant for sonably be included in the denominators used to inhibitor, Km, binding constant for the substrate, evaluate the proportions of treated patients shed- and [S], substrate concentration. Therefore, if the ding resistant virus. Of note, immunocompro- substrate concentration is fixed in a particular mised hosts can have protracted shedding of assay, then the IC50 is proportional to the Ki for influenza viruses for weeks and sometimes months any particular NA of fixed Km. If the substrate (Klimov et al., 1995). Careful scrutiny of viral concentration is much greater than Km, then the isolates from such patients is important. Further- observed IC50 value is substantially increased. The more, different routes of drug administration (i.e. IC50 values for different assays and viruses will inhaled versus oral) provide different drug con- differ considerably depending on the Km values centrations and associated selective pressures for for the particular substrates used. Currently the resistance emergence within the respiratory tract. most widely used substrate is the fluorogenic These differences should be considered in collect- reagent 2%-(4-methylumbelliferyl)-a-D-N-acetyl- ing samples for recovery of potentially resistant neuraminic acid (MUNANA), although a chemi- viruses. For example, orally inhaled zanamivir is luminescent reagent, the 1,2-diotetane derivative predominately deposited in the pharynx and tra- of sialic acid (NA-STAR) may offer greater sensi- cheobronchial tree. Consequently, throat or lower tivity (Buxton et al., 2000). Km values will also respiratory samples (sputum, tracheal aspirates) vary by NA type (A versus B), subtype, and from would be more appropriate than nasal ones in isolate to isolate. Increasing substrate concentra- searching for resistant variants. tions can magnify shifts in susceptibility deter- mined by NA inhibition assays but also increase the background noise and variance of the system. 8. Characterization of resistant variants and Before screening of large numbers of clinical iso- potential for transmission lates, it is essential to validate assay conditions, including optimal substrate concentrations and The probability of transmitting NI-resistant buffer systems. influenza virus person-to-person relative to wild- type influenza could be very low, particularly if resistant variants have compromised viral fitness. 7. Sampling and interpretation of resistance data Thus, broadly based surveillance may underesti- mate the frequency of resistance emergence in Influenza virus infection is an acute event in the individual patients. Therefore, it would be useful, immunocompetent. Virus is cleared completely by although more difficult, to include sampling of the host’s immune system following each infec- patients at higher risk of prolonged replication 154 M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156 and resistance emergence (e.g. children, immuno- low). However, representative isolates and origi- compromised, hospitalized) during treatment. nal samples should be stored for the time that Whenever possible, last day isolates collected on reliable cell-based assays are available. Prelimi- or after day 3 of therapy should be submitted for nary data on the variation in susceptibility of over analysis. Here, the chances of observing resistance 1000 natural influenza isolates collected from should be higher, particularly if resistant strains many regions of the world prior to the introduc- are compromised and not easily transmitted. In tion of NIs is being generated currently with the the absence of a reliable cell-based assay, it would co-operation of the WHO Global Influenza also be useful to sequence the HA from matched Surveillance Network. Susceptibility is being as- pairs of isolates during treatment to observe if sessed by NA inhibition assay for both approved any consistent patterns of mutations in HA NIs. NA gene sequencing will be determined for emerges. The antigenic characterization of viruses samples showing reduced or outlying susceptibil- pre and post drug exposure, particularly those in ity. Results will be scrutinised by the Network which HA mutations affecting the receptor bind- members, and this process will provide informa- ing sites are recognized, is also important. tion of the baseline of NI susceptibility prior to The HA/NA balance required to infect MDCK the introduction of drugs. This assessment will (or other laboratory cell lines) may not reflect that provide data regarding the magnitude of natural for human respiratory tract epithelium. A suitable variation in susceptibility in clinical isolates and human respiratory cell system for phenotypic provide one point of reference for subsequent characterization of NI-resistant variants remains prospective monitoring of resistance emergence. It to be established. Thus growth characteristics in should also provide informed comment about the MDCK cells of any mutant virus compared to different technical approaches to determination of wild-type may not reflect potential growth (infec- NI resistance to ensure that there are accurate tivity and replication) characteristics in the human estimates of resistance. If clinically significant re- respiratory tract. HA mutations can mask poten- sistance emerges, it will be important to rapidly tial defects in growth due to NA instability or disseminate this information to assist planning in reduced activity in vitro, so that growth properties different regions. in cell cultures or eggs may not reflect loss of viral replication fitness in vivo. Infectivity/replicative ability of mutant viruses is currently best assessed 10. Conclusions in vivo in the ferret. This species is potentially the best model for human influenza, given the similar- ity between the two species with regard to recep- 1. There are a number of technical difficulties tor type and the consequent ability to infect associated with determining NI resistance in ferrets without virus adaptation. Furthermore, influenza viruses. Previous cell culture method- correlation has been established between the in ologies used for other viral systems may not be vivo inhibitory effect of NIs in the ferret and virus suitable, and novel assays need to be estab- susceptibility determined by NA inhibition assay, lished. The plaque reduction assay is not suit- but not in MDCK plaque assays (Tisdale, 2000). able as the sole assay for testing NI resistance and currently an NA inhibition assay is the most predictive assay for susceptibility 9. Initial strategy for NI susceptibility monitoring. determination 2. Mutations in viral NA and HA can both contribute to the resistance phenotype, al- No cell culture-based assay (e.g. plaque reduc- though the relative significance of these re- tion, yield reduction, EIA) can currently be rec- mains in vivo to be established. Sequence ommended for reliable assessment of NI analysis of key regions of the HA gene and susceptibility (reviewed in Tisdale, 2000; see be- assessment of antigenic changes in isolates M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156 155

with apparent drug-related changes are neces- neuraminidase assay for the determination of influenza sary at present. virus susceptibility to zanamivir. Anal. Biochem. 280, 291– 300. 3. The relationship between phenotype and geno- Colman, P.M., 1994. Influenza virus neuraminidase: structure, type of NI-resistant viruses remains to be fully antibodies, and inhibitors. Protein Sci. 3 (10), 1687–1696. clarified, although certain NA mutations have Degelau, J., Somani, S.K., Cooper, S.L., Guay, D.R., Cross- been shown to confer decreased enzyme inhibi- ley, K.B., 1992. Amantadine-resistant influenza A in a tion by NI. nursing facility. Arch. Intern. Med. 152 (2), 390–392. 4. Surveillance for NI resistance should be estab- Englund, J.A., Champlin, R.E., Wyde, P.R., Kantarjian, H., Atmar, R.L., Tarrand, J.J., et al., 1998. Common emer- lished globally to reflect the global impact of gence of amantadine and rimantadine resistant influenza A influenza and to reflect the increasing use of viruses in symptomatic immunocompromised adults. Clin. different NA inhibitors. Infect. Dis. 26 (6), 1418–1424. 5. Because of the limited information regarding Gubareva, L.V., Matrosovich, M.N., Brenner, M.K., Bethell, resistance emergence during clinical use, R., Webster, R.G., 1998. Evidence for zanamivir resistance in an immunocompromised child infected with influenza B targeted surveillance in risk populations, in- virus. J. Infect. Dis. 178 (5), 1257–1262. cluding young children, immunocompromised Gubareva, L.V., Kaiser, L., Hayden, F.G., 2000. Influenza hosts, and elderly institutionalized adults, and virus neuraminidase inhibitors. Lancet 355, 827–835. correlations of resistance detection with viral Gubareva, L.V., Kaiser, L., Matrosovich, M.N., Soo-Hoo, Y., replication and clinical outcomes are needed. Hayden, F.G., 2001. Selection of influenza virus mutants in experimentally infected volunteers treated with oseltamivir. In summary, NA inhibition phenotyping of initial J. Infect. Dis. 183, 523–531. (pre-treatment) and final positive virus isolates, Hall, C.B., Dolin, R., Gala, C.L., Markovitz, D.M., Zhang, supported by NA and HA sequencing, currently Y.Q., Madore, P.H., et al., 1987. Children with influenza A provides a reliable and reasonably comprehensive infection: treatment with rimantadine. Pediatrics 80 (2), approach to identification of NI resistant clinical 275–282. isolates. Hay, A.J., 1992. The action of adamantanamines against influenza A viruses: inhibition of the M2 ion channel protein. Semin. Virol. 3, 21–30. Hayden, F.G., 1996. Amantadine and rimantadine-clinical as- References pects. In: Richman, D.D. (Ed.), Antiviral Drug Resistance. Wiley, New York, pp. 59–77. Barnett, J.M., Cadman, A., Burell, F.M., Madar, S.H., Lewis, Hayden, F.G., Belshe, R.B., Clover, R.D., Hay, A.J., Oakes, A.P., Tisdale, M., et al., 1999. In vitro selection and M.G., Soo, W., 1989. Emergence and apparent transmis- characterisation of influenza B/Beijing/1/87 isolates with sion of rimantadine-resistant influenza A virus in families. altered susceptibility to zanamivir. Virology 265, 286–295. N. Engl. J. Med. 321 (25), 1696–1702. Barnett, J., Cadman, A., Gor, D., Dempsey, M., Walters, M., Hayden, F.G., Sperber, S.J., Belshe, R.B., Clover, R.D., Hay, Candlin, A., et al., 2000. Zanamivir susceptibility monitor- A.J., Pyke, S., 1991. Recovery of drug-resistant influenza A ing and characterization of influenza virus clinical isolates virus during therapeutic use of rimantadine. Antimicrob. obtained during phase II clinical efficacy studies. Antimi- Agents Chemother. 35 (9), 1741–1747. crob. Agents Chemother. 44 (1), 78–87. Klimov, A.I., Rocha, E., Hayden, F.G., Shult, P.A., Roumil- Belshe, R.B., Smith, M.H., Hall, C.B., Betts, R., Hay, A.J., lat, L.F., Cox, N.J., 1995. Prolonged shedding of aman- 1988. Genetic basis of resistance to rimantadine emerging tadine-resistant influenzae A viruses by immunodeficient during treatment of influenza virus infection. J. Virol. 62, patients: detection by polymerase chain reaction-restriction 1508–1512. analysis. J. Infect. Dis. 172 (5), 1352–1355. Blick, T.J., Sahasrabudhe, A., McDonald, M., Owens, I.J., Mast, E.E., Harmon, M.W., Gravenstein, S., Wu, S.P., Arden, Morley, P.J., Fenton, R.J., et al., 1998. The interaction of N.H., Circo, R., et al., 1991. Emergence and possible neuraminidase and hemagglutinin mutations in influenza transmission of amantadine-resistant viruses during nurs- virus in resistance to 4-guanidino-Neu5Ac2en. Virology ing home outbreaks of influenza A (H3N2). Am. J. Epi- 246 (1), 95–103. demiol. 134 (9), 988–997. Boivin, G., Goyette, N., Hardy, I., Aoki, F.Y., Wagner, A., McKimm-Breschkin, J.L., 2000. Resistance of influenza Trottier, S., 2000. Rapid antiviral effect of inhaled viruses to neuraminidase inhibitors — a review. Antivir. zanamivir in the treatment of naturally occurring influenza Res. 47, 1–17. in otherwise healthy adults. J. Infect. Dis. 181, 1471–1474. Mendel, D.B., Sidwell, R.W., 1998. Influenza virus resistance Buxton, R.C., Juo, R.R., Voyta, J.C., Tisdale, M., Bethell, to neuraminidase inhibitors. Drug Resist. Updates 1, 184– R.C., 2000. Development of a sensitive chemiluminescent 189. 156 M. Zambon, F. G. Hayden / Anti6iral Research 49 (2001) 147–156

NICE, 1999. National Institute for Clinical Excellence. Treanor, J.J., Hayden, F.G., Vrooman, P.S., Barbarash, Zanamivir (Relenza) in the management and treatment of R.A., Bettis, R., Riff, D., et al., 2000. Efficacy and safety influenza. Guidance 1999. of the oral neuraminidase inhibitor oseltamivir in treating Penn, C.R., Barnett, J., Bethell, R., Fenton, R.J., Gearing, acute influenza. JAMA 283 (8), 1016–1024. L., Healy, N. et al., 1996. Selection of influenza virus Tisdale, M., 2000. Monitoring of viral susceptibility: new with reduced sensitivity in vitro to the neuraminidase challenges with the development of influenza NA in- inhibitor GG167 (4-guanidino-Neu5Ac2en): changes in hibitors. Rev. Med. Virol. 10, 45–55. haemagglutinin may compensate for loss of neu- WHO, 1996. WHO influenza surveillance, Weekly Epidemio- raminidase activity. In: Brown, L.E., Hampson, A.W., logical Record 22; 71 (47): 353–357. Webster, R.G., editors. Options For The Control of Infl- Woods, J.M., Bethell, R.C., Coates, J.A., Healy, N., Hiscox, uenza III. Elsevier Science BV, Amsterdam, The Nether- S.A., Pearson, B.A., et al., 1993. 4-Guanidino-2,4- lands, pp. 735–740. dideoxy-2,3-dehydro-N-acetylneuraminic acid is a highly Richman DD, 1996. Issues and Challenges. In: Richman effective inhibitor both of the sialidase (neuraminidase) DD, editor. Antiviral Drug Resistance. Chichester: Wiley, and of growth of a wide range of influenza A and B p. 6. viruses in vitro. Antimicrob. Agents Chemother. 37 (7), Sidwell, R.W., Huffman, J.H., Barnard, D.L., Bailey, K.W., 1473–1479. Wong, M.H., Morrison, A., et al., 1998. Inhibition of Ziegler, T., Hemphill, M.L., Ziegler, M.L., Perez-Oronoz, influenza virus infections in mice by GS4104, an orally G., Klimov, A.I., Hampson, A.W., et al., 1999. Low effective influenza virus neuraminidase inhibitor. Antiviral incidence of rimantadine resistance in field isolates of Res. 37, 107–120. influenza A viruses. J. Infect. Dis. 180 (4), 935–939.

.