The combination of automated purification of nucleic acids with real-time PCR should enable even more rapid identificati
Reviews in Medical Virology
REVIEW
Rev. Med. Virol. 2002; 12: 375–389. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rmv.370
Molecular diagnosis of influenza Joanna S. Ellis* and Maria C. Zambon Respiratory Virus Unit, Enteric, Respiratory and Neurological Virus Laboratory, Public Health Laboratory Service, Central Public Health Laboratory, 61 Colindale Avenue, Colindale, London NW9 5HT, UK
SUMMARY The past decade has seen tremendous developments in molecular diagnostic techniques. In particular, the development of PCR technology has enabled rapid and sensitive viral diagnostic tests to influence patient management. Molecular methods used directly on clinical material have an important role to play in the diagnosis and surveillance of influenza viruses. Molecular diagnostic tests that allow timely and accurate detection of influenza are already implemented in many laboratories. The combination of automated purification of nucleic acids with real-time PCR should enable even more rapid identification of viral pathogens such as influenza viruses in clinical material. The recent development of DNA microarrays to identify either multiple gene targets from a single pathogen, or multiple pathogens in a single sample has the capacity to transform influenza diagnosis. While molecular methods will not replace cell culture for the provision of virus isolates for antigenic characterisation, they remain invaluable in assisting our understanding of the epidemiology of influenza viruses. Copyright # 2002 John Wiley & Sons, Ltd. Accepted: 7 August 2002
INTRODUCTION Influenza viruses continue to be a major cause of respiratory tract infection, resulting in significant morbidity, mortality and financial burden. Each year, increased hospitalisation rates and excess deaths are attributable to influenza infections. Although influenza for most people is a mild illness, which is resolved in 1–2 weeks, complications associated with influenza infection can occur in both the upper and lower respiratory tract, some of which may be fatal [1]. The potential for developing complications is higher in certain risk groups, such as the elderly and individuals with chronic medical conditions. An accurate diagnosis of influenza by a physician is difficult since several different pathogens can produce respiratory ill*Corresponding author: Dr J. S. Ellis, Respiratory Virus Unit, Enteric, Respiratory and Neurological Virus Laboratory, Public Health Laboratory Service, Central Public Health Laboratory, 61 Colindale Avenue, Colindale, London NW9 5HT, UK. E-mail:
[email protected] Abbreviations used BDNA, branched chain DNA; EIA, enzyme immunoassay; HI, haemagglutination inhibition; HMA, heteroduplex mobility assay; IF, immunofluorescence; ILI, influenza-like illness; LCR, ligase chain reaction; M, matrix; NASBA, nucleic acid sequence-based amplification; NP, nucleoprotein; NS, non-structural; PIV, parainfluenza; RFLP, restriction fragment length polymorphism; WHO, World Health Organisation.
Copyright # 2002 John Wiley & Sons, Ltd.
nesses with similar clinical symptoms. Consequently, there is a requirement for sensitive and rapid diagnostic techniques to verify the clinical diagnosis of influenza and improve the quality of surveillance systems. Moreover, the development of specific anti-influenza neuraminidase inhibitors has increased the potential for rapid and accurate diagnostic tests for influenza viruses to contribute to the management of patients. A number of laboratory methods for the diagnosis of influenza are currently available (Table 1). Each of these methods has advantages and disadvantages, and some, or all, of these factors may influence the method of choice. Although molecular technology has transformed the diagnosis of a number of diseases caused by RNA viruses, for example HIV [2], the application of molecular methods to the detection of respiratory pathogens is still comparatively new and expanding rapidly. MOLECULAR METHODS FOR THE DETECTION OF INFLUENZA VIRUSES
Choice of molecular assay A number of molecular methods can be employed for the detection of influenza viruses, the majority of which are based on PCR methodology (Table 2). When selecting which assay to use, there are a
376
J. S. Ellis and M. C. Zambon
Table 1. Comparison of the properties of diagnostic methods for influenza virus detection Method Sample required
Cost per test Speeda (£)
Through putb
Advantages
Culture Nasopharyngeal aspirate Nose and throat swab Bronchoalveolar lavage IF Nasopharyngeal aspirate Nose and throat swab Bronchoalveolar lavage
£10–£20
3–7 days
Low
Whole virus Requires measured infectious virus Virus Highly skilled recoverable Time required
£10
2 h–1 day
Medium
Rapid
EIA
Nasopharyngeal £5–£30 aspirate Nose and throat swab Bronchoalveolar lavage RT-PCR Nasopharyngeal £30–£50 aspirate Nose and throat swab Bronchoalveolar lavage Post-mortem tissue Serology Serum £5 HI/CFT
15 min–1 day High
1–2 days
High
2 days
High
Rapid Low skill Can be ‘near patient’ testing
Disadvantages
Requires intact cells Highly skilled No virus recoverable Specialised equipment Cost No virus recoverable
Sensitive Allows further molecular analysis
Cost Highly skilled No virus recoverable Specialised equipment/ laboratory Sensitive and Retrospective specific Paired samples needed
a
Speed with which results are available from the time the specimen is taken. Number of specimens handled in unit time.
b
number of factors that should be taken into consideration. These include the requirement for qualitative, semi-quantitative or quantitative data, and the nature and number of samples to be analysed. In addition, the available time and resources of the laboratory where the work is to be performed and the skill of the staff involved must also be considered.
Hybridisation Molecular hybridisation methods have been used extensively in basic and applied virology because of their technical flexibility and high specificity. Copyright # 2002 John Wiley & Sons, Ltd.
Using these techniques, DNA and RNA viruses have been detected directly in clinical specimens. Influenza A virus RNA has been detected in nasopharyngeal swabs by molecular hybridisation, with a sensitivity of 72% [3]. Hybridisation was more sensitive than both immunofluorescence (IF) and culture, but less sensitive than EIA, which detected influenza virus in 86% of swabs. Molecular hybridisation, however, has not been widely used for the detection of influenza viral RNA in clinical material, largely due to the development and optimisation of more sensitive and less timeconsuming techniques. Rev. Med. Virol. 2002; 12: 375–389.
Molecular diagnosis of influenza
377
Table 2. Properties of molecular methods suitable for detection of influenza viruses Method
Advantages
Disadvantages
Hybridisation
High specificity Flexible formats Inexpensive Sensitive and specific Allows further product analysis Sensitive and specific Tests >1 target per assay Allows further product analysis Sensitive and specific High throughput Not limited by product size Can be multiplexed Sensitive and specific Rapid Can be quantitative Can be multiplexed Sensitive and specific Can be quantitative
Sensitivity Time-consuming
PCR Multiplex PCR
PCR-EIA
Real-time PCR
NASBA
Microarrays
Sensitive and specific Tests many targets per assay
PCR The development of PCR analysis in 1985 made possible the diagnosis of virus infection through sensitive detection of specific viral nucleic acids [4]. PCR techniques have been developed for the specific detection and subtyping of influenza viruses to obtain rapid diagnostic results. In these assays, purified influenza viral RNA from culture fluids or clinical specimens is first reverse transcribed to cDNA, by either avian myeloblastosis virus or moloney murine leukaemia virus reverse transcriptase, using random hexanucleotides, universal primer complementary to the 30 end of all influenza vRNAs, or a sequence specific primer. The use of random hexanucleotides or a universal primer, instead of target sequence specific primers has the advantage that the cDNA from both viral genomic RNA and mRNA transcripts is synthesized, thereby increasing the number of target regions that could be amplified by PCR. A single round of amplification may be used and the specificity of the reaction confirmed by hybridisation with product specific probes. Alternatively, Copyright # 2002 John Wiley & Sons, Ltd.
Qualitative, not usually quantitative Requires extensive optimisation to ensure no false negatives or primer competition May be limited by product size Does not allow further product analysis Requires additional evaluation/validation before use Requires specialised equipment Product analysis not always feasible
Three enzymes used Time-consuming RNA product analysis less feasible Requires specialised equipment and extensive development Does not allow further product analysis
nested-primer sets may be utilised to amplify the target region. The choice of gene target is influenced by the prospective application of the assay. For typespecific diagnosis of influenza A, B or C infection, internal genes such as nucleoprotein (NP) and matrix (M) genes are usually chosen, since these are highly conserved within influenza types. Where information on the subtype of influenza A is required, then the genes encoding the surface antigens are targeted. The use of multisegment PCR using primers complementary to the conserved 13 nucleotides at the 50 terminus and 12 nucleotides at the 30 terminus allows the detection of all segments in a single reaction [5]. Moreover, primer sets based on the 15 and 21 nucleotide terminal sequences enable the specific amplification of each of the eight RNA segments and subsequent analysis of all 15 HA and 9 NA subtypes of influenza A virus [6]. Reports on the applications of PCR techniques to the detection of influenza A and B viruses first appeared in the early 1990s. Initial reports described Rev. Med. Virol. 2002; 12: 375–389.
378 the amplification of influenza A H1 nucleic acid from both nasopharyngeal lavages [7] and nasopharyngeal aspirates [8], and influenza A H1, H3 and influenza B HA specific sequences from throat swabs [9,10]. In a study by Zhang and Evans [11] the M genes of influenza A and B viruses and the influenza C HA gene were targets for type-specific nested-primer sets. Subtype-specific primers targeted conserved sequences within the three HA or two NA subtypes of different human influenza isolates. Although the assays were shown to be both sensitive and specific for the typing and subtyping of cultured influenza viruses, their application to the analysis of viruses directly from clinical material was not assessed. Further studies reported the detection of influenza A, B and C viruses in respiratory secretions, using primers targeting specific sequences in the non-structural (NS), or M genes [12–14]. In these initial studies, a small number of respiratory samples were used to evaluate the applicability of the RT-PCR assays for the diagnosis of influenza in clinical specimens. Although early assays used different extraction procedures, gene targets and methods for discrimination of products, the sensitivity of PCR for the detection of influenza viruses was shown to be at least equivalent to that of culture. The majority of RT-PCR assays described to date have utilised primers deduced from the nucleotide sequences of human influenza viral genes. The occurrence of zoonotic events in Hong Kong and Mainland China caused by influenza A H5N1 and H9N2 avian viruses have highlighted the additional requirement for rapid tests to detect viruses originating from non-human hosts. Recently, a single-tube RT-PCR assay for the specific detection of influenza A viruses from multiple species has been reported [15]. A primer set based on conserved regions of the M gene was shown to detect a panel of 25 genetically diverse virus isolates obtained from birds, humans, pigs, horses and seals which included all known subtypes of influenza A virus. To detect and partially characterise influenza A viruses from different animal species, a combined RT-PCR heteroduplex mobility assay (HMA) approach has been described [16]. An M gene-based RT-PCR using nested primer pairs was shown to be specific for the detection of human, avian and swine influenza A viruses, of 15 different subtypes. PCR amplicons were then analysed by HMA to determine the Copyright # 2002 John Wiley & Sons, Ltd.
J. S. Ellis and M. C. Zambon host of origin of the M gene segment. This methodology offers a rapid and sensitive means of screening for novel or unusual influenza viruses. The segmented nature of the genome of influenza A, B and C viruses makes reassortment among viruses an important mechanism for generating genetic diversity. Reassortment between influenza A viruses of different subtypes is of particular importance because of its role in the generation of new pandemic strains in humans. Hence, there is a need for the development of rapid molecular assays that provide subtype information on both the HA and NA of influenza A viruses. This is highlighted by the recent emergence of influenza A H1N2 reassortant viruses in several countries during 2001–2002, following reassortment between circulating influenza A H1N1 and H3N2 viruses [17]. The majority of existing molecular diagnostic tests for influenza A viruses rely on the detection and typing of the HA gene alone, with only a few laboratories undertaking antigenic or genetic analysis of the NA. As yet, PCR assays targeting specific NA subtype sequences have only been described for the N1 and N2 genes of influenza A viruses [11,18].
PCR-EIA A modification of the PCR technique, PCR-enzyme immunoassay (PCR-EIA) has been described [19–22]. A single round RT-PCR is performed and the resulting amplicons identified by hybridisation in solution to a biotinylated RNA probe and detected in an EIA. The potential efficiency of influenza A, M-gene specific PCR-EIA for use in clinical diagnosis was evaluated in a study by Cherian [19]. A total of 90 nasal wash specimens were obtained daily over a 10-day period from nine human volunteers infected with influenza A virus. The PCR-EIA and cell culture had equivalent sensitivities during the first 4 days following infection. However, PCR-EIA was substantially more sensitive than culture later in the course of infection. From days 5 to 9, PCR-EIA could detect influenza A virus RNA in 38% of samples, whereas only 9% were culture positive during that period of time. The results demonstrated that clinical symptoms and shedding of viral RNA might persist after virus becomes undetectable by traditional cultivation methods. The clinical importance of the detection of viral RNA sequences in culture negative samples remains to Rev. Med. Virol. 2002; 12: 375–389.
Molecular diagnosis of influenza
379
be established, although increased sensitivity of this analytical method has several advantages, including extending the window for detection of virus and diagnosis of infection.
Multiplex PCR The term multiplex PCR refers to the inclusion of more than one primer set in an amplification reaction, to detect the presence of more than one gene or genome segment in a single pathogen, or in more than one pathogen. Several multiplex PCR assays have been described for the detection of respiratory pathogens in clinical material (Table 3). Multiplex PCR assays have been designed to type and subtype influenza A H1, H3 and influenza B virus in clinical specimens [18,23]. A multiplex PCR assay using nested primer sets targeting conserved regions of the HA genes of influenza A H1, H3 and influenza B viruses has been used in a prospective surveillance of influenza in England during the 1995–1996 winter season [23]. A total of 619 combined nose and throat swabs from patients with an influenza-like illness (ILI) were analysed by cell culture and multiplex PCR. Of these,
39.7% were positive by multiplex PCR, compared to 32.3% that yielded influenza viruses by cell culture. The benefits of multiplex RT-PCR compared with cell culture for detection were most evident during the peak period when influenza virus was circulating in the community. During this period, 38.9% of the samples received were positive by cell culture, whereas 57.5% were positive for influenza virus by multiplex PCR, indicating the improved positive predictive value of multiplex PCR compared with cell culture. Since infection with respiratory pathogen(s) other than influenza viruses can result in clinical symptoms clinically indistinguishable from true influenza, samples negative for influenza virus from patients with ILI may contain these pathogens. The multiplex PCR assay described above has been modified to detect and subtype RSV A and B, in addition to influenza A and B viruses, in a single reaction [24]. In this study, an excellent (100%) correlation between multiplex PCR and the results of influenza culture was demonstrated. The test was also able to detect mixed viral infections in both simulated specimens and clinical samples.
Table 3. Multiplex RT-PCR assays described for detecting human respiratory pathogens Target
Number of pathogens detected
Type Method of of product PCRa identification
Sensitivity
Reference
Influenza A (H1, H3) and B Influenza A and RSV
3
S
25–100 pfu
[18]
2
S
2 pfu
[20]
Influenza A (H1, H3) and B Influenza A (H1, H3), B and C Influenza A (H1, H3), B and RSV A and B Influenza A and B, RSV A and B, HPIV-1, -2 and -3 (Hexaplex assay) Influenza A and B, RSV, HPIV-1, -3, enterovirus, adenovirus, Mycoplasma pneumoniae and Chlamydia pneumoniae
3
N
1–5 pfu
[23]
4
S
Not reported
[49]
5
N
1 pfu
[24]
7
S
5–10 genome copies
[22]
9
S
100 genome copies
[21]
a
Size/Gel electrophoresis Size/Gel electrophoresis Size/Gel electrophoresis Size/Gel electrophoresis Size/Gel electrophoresis Probe/Enzyme hybridisation assay Probe/PCRELISA
Number of rounds of amplification; S, single, one round of amplification, N, nested, two rounds of amplification.
Copyright # 2002 John Wiley & Sons, Ltd.
Rev. Med. Virol. 2002; 12: 375–389.
380 The use of multiplex RT-PCR enzyme hybridisation assays has allowed the detection of a wide range of respiratory pathogens in a single test, since discrimination of the products is not limited by size differences detectable following gel electrophoresis. The Hexaplex assay (Prodesse, Milwaukee) includes primers specific for the M and NS genes of influenza A and B viruses respectively, as well as other primer pairs for the detection and quantification of RSV A and B, human parainfluenza (PIV) type-1, 2 and 3 RNA in clinical specimens [22]. A hot start RT-PCR EIA protocol for the simultaneous detection of nine respiratory pathogens including influenza A and B viruses, PIV-1, PIV-3, RSV, enteroviruses, adenovirus, Chlamydia pneumoniae and Mycoplasma pneumoniae has been developed. This assay has been evaluated using 1118 nasopharyngeal aspirates collected from hospitalised children over a number of influenza seasons [21,25]. Of these, 395 (35%) were positive for one of the nine pathogens. Over the period of the study, 20% of these were positive for influenza A virus and 2.8% were positive for influenza B virus. Seasonal variations in the rates of detection of the different infectious agents were seen. The multiplex RT-PCR showed excellent levels of concordance (98%) for influenza A virus detection with data obtained by commercially available enzyme immunoassay.
NASBA In the past few years, amplification procedures other than PCR have been developed. These include branched chain DNA (bDNA) assay [26], ligase chain reaction (LCR) assays [27], and nucleic acid sequence-based amplification (NASBA). Of these assays, NASBA has been most widely applied for the detection of viral nucleic acid. In the NASBA system amplification is carried out isothermally using three enzymes simultaneously (avian myeloblastosis virus reverse transcriptase, T7 RNA polymerase and RNase H) with two specially designed DNA oligonucleotide primers. The technique is particularly suited to the detection of RNA viruses, since RNA polymerase amplifies RNA without conversion to complementary DNA [28]. NASBA has been applied successfully to a number of RNA templates including HIV-1 [29], HCV [30], rhinoviruses [31] and enteroviruses [32]. Interestingly, a NASBA technique has very recently been developed that allows the detection Copyright # 2002 John Wiley & Sons, Ltd.
J. S. Ellis and M. C. Zambon of avian influenza A subtype H5 isolates of the Eurasian lineage from allantoic fluid harvested from inoculated chick embryos [33]. Both generic primers (for the amplification of both highly pathogenic and low pathogenic H5 HA sequences), and primers specific for pathogenic H5 HA sequences were designed and evaluated for the detection of H5 HA sequences and the differentiation of highly pathogenic H5 sequences from low pathogenic H5 sequences. However, NASBA technology has yet to be applied to the detection of human influenza viruses in respiratory specimens. QUALITATIVE VERSUS QUANTITATIVE ASSAYS The most sensitive molecular techniques use an enzymatic step to amplify target nucleic acid before detection of the specific sequence. Typically, the logarithmic amplification of the target sequence results in the amount of DNA product at the end of the PCR having no correlation to the number of target copies present in the original specimen. The qualitative nature of PCR-based applications limited, in the early stages, the use of this methodology to those conditions where only the presence or absence of a specific target sequence was to be assayed. However, many applications require quantification of the number of target molecules in the original specimen. Both PCR-based and non-PCR strategies, such as realtime PCR and NASBA, respectively, have been adapted for quantitation of nucleic acid products, using either internal or external standards. Quantitative molecular assays have been used primarily to determine the viral load in blood specimens for the diagnosis of HIV, HCV and hepatitis B virus infections [34–36]. Quantitation of CMV DNA molecules is required for successful monitoring of this infection in immunocompromised individuals and during therapy. Where traditional diagnostic methods are insensitive, our understanding of the natural history of viral disease has continued to be improved through the use of molecular techniques. In particular, quantitative PCR-based studies have had an important role in the analysis of the natural history of HIV and HCV infections, and subsequently in the development of effective treatment strategies. During antiviral treatments, quantitative techniques can supply information on both efficacy of therapies and selection of drug-resistant Rev. Med. Virol. 2002; 12: 375–389.
Molecular diagnosis of influenza viral variants in real-time, as demonstrated in HIV-1 infection [37,38]. Although quantitative PCR assays have yet to be widely employed in the diagnosis of influenza infections, and the development of such assays is technically demanding, the potential value of such assays in monitoring efficacy of anti-influenza drug treatments and in the monitoring of immunosuppressed patients is apparent. ANALYSIS OF MOLECULAR PRODUCTS One advantage of PCR-based diagnostic assays is that subsequent sequence information can be
381 determined from the assay products, in a variety of ways (Figure 1). Post-amplification methods for analysing sequence variation in PCR amplicons include restriction fragment length polymorphism analysis (RFLP), HMA and nucleotide sequencing.
PCR-RFLP A combination of RT-PCR and enzyme digestion of HA gene amplicons (PCR-restriction assay) has been used to rapidly differentiate influenza A H3 genetic variants that are antigenically similar [39,40]. This technique may also be utilised to dif-
Figure 1. PCR-based amplification methods and product analysis
Copyright # 2002 John Wiley & Sons, Ltd.
Rev. Med. Virol. 2002; 12: 375–389.
382 ferentiate between vaccine strains and currently circulating strains [38] and to assess the gene composition of candidate vaccine strains [41,42]. Amantadine-resistant influenza A viruses have been detected in nasopharyngeal swabs by PCRRFLP [43]. PCR-restriction assays have most recently been used to rapidly genotype and monitor the internal genes of human influenza H1N1, H3N2 and H5N1 influenza A viruses [44]. A possible drawback of the RFLP technique is that mutations in the nucleotide sequence of the gene of interest may lead to the loss or generation of a restriction site. The high mutability of RNA genomes such as influenza viral genes increases the possibility of this occurring.
Heteroduplex mobility assays Heteroduplex formation between amplicons has been used to subtype viral genomes, to screen for genome variants, to detect mutations and to measure diversity within and between viral genome populations [45]. The initial application of HMA in virology was to subtype HIV-1 [46,47]. This methodology has since been applied to the analysis of many RNA viral genomes, including those of influenza A and B viruses. By combining a HAspecific multiplex RT-PCR with HMA, variant strains of influenza A and B viruses can be differentiated [48,49]. Direct amplification of internal genes from clinical material can also be coupled with HMA to allow rapid species identification of the origin of influenza viruses [16].
Nucleotide sequence analysis Variation among amplified fragments indicated by RFLP or HMA can be confirmed by nucleotide sequencing of derived clones or the PCR product itself. In many laboratories sequence analysis of PCR amplicons is routinely performed, particularly on HA gene products where an association between sequence changes with genetic drift is studied [50,51]. Sequence analysis of influenza gene segments amplified by PCR may also be undertaken to provide information on genetic reassortment between influenza A viruses [52–55]. Antigenic drift in influenza viruses results from the progressive accumulation of mutations, including substitutions, deletions and insertions in the influenza viral genes. These mutations arise due to replication of the influenza virus genome by a viral RNA polymerase that lacks proofreadCopyright # 2002 John Wiley & Sons, Ltd.
J. S. Ellis and M. C. Zambon ing activity. Because of this genetic variation, the target regions in newly emerging strains complementary to the primers or probes used in an assay must be regularly analysed by nucleotide sequencing to check for sequence mismatches between the primer or probe, and target sequences. In the event of sequence mismatches, the primers or probes should be updated in order to avoid false negative assay results. APPLICATIONS OF MOLECULAR METHODS
Diagnosis A limited number of studies have compared laboratory diagnostic techniques for influenza viruses, including RT-PCR, with clinical diagnosis. In a study that aimed to predict influenza infections during epidemics by use of a clinical case definition, combined nasal and pharyngeal swab specimens were collected from 100 patients presenting with ILI of
